Adaptive motion vector decision method and device for digital image stabilizer system
First Claim
1. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
- local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field, and sequentially generating respective local motion vectors of said local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field and including the respective instability of each of said local motion vectors in each most recent field respective to a field just previous;
field motion vector generation means for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, generating a respective stability weight signal that is of decreased value with increased instability, generating a respective output weight signal for each said local motion vector by additively combining its respective isolation and respective stability weights and generating a respective field motion vector by weighted averaging of the local motion vectors within each field, said weighted averaging including a multiplying of each of said local motion vectors within each field by a corresponding one of said output weight signals to form a product that selectively is included or is not included at all within a vector sum of weighted local motion vectors within each field, which vector sum is divided by a sum of the weights just of said weighted local motion vectors included in said vector sum in order to generate said respective field motion vector; and
accumulated motion vector generation means for receiving said field motion vector and generating said accumulated motion vector by accumulating said received field motion vector into an attenuated response to the previous accumulated motion vector supplied byattenuator means included within said accumulated motion vector generation means.
1 Assignment
0 Petitions
Accused Products
Abstract
A digital image stabilization device receives correlation data generated from a plurality of local motion estimation areas and adaptively decides a motion vector according to irregular image environment. To do this, statistical variables and local motion vector are generated to check reliability by checking correlation values generated from each local motion estimation area. Then, it is determined that there is irregular image data deteriorating reliability of local motion vector, by analyzing statistical variables. Here, the irregular image environment can be in an image signal with low contrast, an image signal with a moving object, an image signal with repeated motion, etc. Also, isolativity and stability are obtained by analyzing local motion vector and a weight signal is generated with a predefined rate according to a computed result. Then, the output of the weight signal is controlled, according to a result of the reliability determination signal, where if there is irregular image data, the output of weight signal is blocked. And, a field motion vector is generated by multiplying a weight signal by a corresponding local motion vector and then averaging them. Finally, an accumulated motion vector is generated by attenuating a previous accumulated motion vector and adding the attenuated previous accumulated motion vector into the field motion vector.
-
Citations
70 Claims
-
1. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field, and sequentially generating respective local motion vectors of said local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field and including the respective instability of each of said local motion vectors in each most recent field respective to a field just previous; field motion vector generation means for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, generating a respective stability weight signal that is of decreased value with increased instability, generating a respective output weight signal for each said local motion vector by additively combining its respective isolation and respective stability weights and generating a respective field motion vector by weighted averaging of the local motion vectors within each field, said weighted averaging including a multiplying of each of said local motion vectors within each field by a corresponding one of said output weight signals to form a product that selectively is included or is not included at all within a vector sum of weighted local motion vectors within each field, which vector sum is divided by a sum of the weights just of said weighted local motion vectors included in said vector sum in order to generate said respective field motion vector; and accumulated motion vector generation means for receiving said field motion vector and generating said accumulated motion vector by accumulating said received field motion vector into an attenuated response to the previous accumulated motion vector supplied by attenuator means included within said accumulated motion vector generation means. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
-
-
9. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field, sequentially generating respective local motion vectors of said local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, which respective isolativity is generated as the magnitude of the result of differentially comparing said local motion vector with an average of the local motion vectors for the same image field, said statistical variables including the respective instability of each of said local motion vectors from a corresponding local motion vector within the previous field, which respective instability is generated as the magnitude of the result of differentially comparing said local motion vector with the corresponding local motion vector within the previous field, and said statistical variables also including other statistical variables besides said isolativities and instabilities; reliability determination circuitry receiving said other statistical variables sequentially supplied from said local motion estimation means, determining from said other statistical variables whether an abnormal condition deteriorating reliability of local motion vector occurs or not, and generating an abnormal-condition signal whenever a result of such determinations is indicative that an abnormal condition occurs; weight signal generation means for receiving for each field said local motion vectors sequentially supplied from said local motion vector estimation means and average local motion vectors generated in fields, generating a respective isolation weight signal for each local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, generating a respective stability weight signal that is of decreased value with increased instability, supplying an output weight signal dependent on the sum of said isolation weight signal and said stability weight signal as long as said reliability determination circuitry does not generate an abnormal-condition signal, and supplying an output weight signal of zero value whenever said reliability determination circuitry generates an abnormal-condition signal; field motion decision means for sequentially receiving said output weight signals and said local motion vectors, and applying said output weight signals to corresponding received local motion vectors of local motion estimation areas, as a part of generating a respective field motion vector for each field by performing a weighted averaging of said local motion vectors within that field; and accumulated motion vector generation means for receiving said field motion vector and generating said accumulated motion vector by accumulating said received field motion vector into an attenuated response to the previous accumulated motion vector supplied by attenuator means included within said accumulated motion vector generation means. - View Dependent Claims (10)
-
-
11. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector said image stabilization device comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching, patterns of said edge data of current field and previous field, sequentially generating respective local motion vectors of said local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, which respective isolativity is generated as the magnitude of the result of differentially comparing said local motion vector with an average of the local motion vectors for the same image field, said statistical variables including the respective instability of each of said local motion vectors from a corresponding local motion vector within the previous field, which respective instability is generated as the magnitude of the result of differentially comparing said local motion vector with the corresponding local motion vector within the previous field, and said statistical variables also including other statistical variables besides said isolativities and instabilities; reliability determination circuitry receiving said other statistical variables sequentially supplied from said local motion estimation means, determining from said other statistical variables whether an abnormal condition deteriorating reliability of local motion vector occurs or not, and generating an abnormal-condition signal whenever a result of such determinations is indicative that an abnormal condition occurs; weight signal generation means for receiving for each field said local motion vectors sequentially supplied from said local motion vector estimation means and average local motion vectors generated in fields, generating a respective isolation weight signal for each local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, generating a respective stability weight signal that is of decreased value with increased instability, generating a respective output weight signal for each said local motion vector by averaging its respective isolation and respective stability weights as long as said reliability determination circuitry does not generate said abnormal-condition signal, and supplying an output weight signal of zero value whenever said reliability determination circuitry generates an abnormal-condition signal, field motion decision means for sequentially receiving said output weight signals and said local motion vectors, and applying said output weight signals to corresponding received local motion vectors of local motion estimation areas, as a part of generating a respective field motion vector for each field by performing a weighted averaging of said local motion vectors within that field; and accumulated motion vector generation means for receiving said field motion vector and generating said accumulated motion vector by accumulating said received field motion vector into an attenuated response to the previous accumulated motion vector supplied by attenuator means included within said accumulated motion vector generation means. - View Dependent Claims (12, 13)
-
-
14. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field, sequentially generating respective local motion vectors of said local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, which respective isolativity is generated as the magnitude of the result of differentially comparing said local motion vector with an average of the local motion vectors for the same image field, said statistical variables including the respective instability of each of said local motion vectors from a corresponding local motion vector within the previous field, which respective instability is generated as the magnitude of the result of differentially comparing said local motion vector with the corresponding local motion vector within the previous field, and said statistical variables also including other statistical variables besides said isolativities and instabilities, reliability determination circuitry receiving said other statistical variables sequentially supplied from said local motion estimation means, determining from said other statistical variables whether an abnormal condition deteriorating reliability of local motion vector occurs or not, and generating an abnormal-condition signal whenever a result of such determinations is indicative that an abnormal condition occurs; weight signal generation means for receiving for each field said local motion vectors sequentially supplied from said local motion vector estimation means and average local motion vectors generated in fields generating a respective isolation weight signal for each local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, generating a respective stability weight signal that is of decreased value with increased instability supplying an output weight signal dependent on said isolation weight signal and on said stability weight signal as long as said reliability determination circuitry does not generate an abnormal-condition signal and supplying an output weight signal of zero value whenever said reliability determination circuitry generates an abnormal-condition signal; field motion decision means for sequentially receiving said output weight signals and said local motion vectors, and applying said output weight signals to corresponding received local motion vectors of local motion estimation areas, as a part of generating a respective field motion vector for each field by performing a weighted averaging of said local motion vectors within that field; panning identification means for receiving said field motion vectors, regarding generation of field motion vectors having substantially identical values over a predetermined number of fields as resulting from deliberate panning, and generating an indication of the occurrence of deliberate panning; means for determining whether or not said received accumulated motion vector tends to leave a range in which undesired motion of said image data stored in said memory can be corrected for; and accumulated motion vector generation means for receiving said field motion vector and generating said accumulated motion vector by accumulating said received field motion vector into an attenuated response to the previous accumulated motion vector supplied by attenuator means included within said accumulated motion vector generation means, for multiplying said previous accumulated motion vector by a first factor smaller than one to generate said attenuated response when there is no indication of the occurrence of deliberate panning generated and it is determined said received accumulated motion vector does not tend to leave said range in which undesired motion of said image data stored in said memory can be corrected for, for multiplying said previous accumulated motion vector by a second factor smaller than said first factor to generate said attenuated response whenever said indication of the occurrence of deliberate panning, is generated, and for multiplying said previous accumulated motion vector by said second factor whenever it is determined said received accumulated motion vector tends to leave said range. - View Dependent Claims (15)
-
-
16. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a compensated accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field and sequentially generating respective local motion vectors of said local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, said statistical variables including the respective instability of each of said local motion vectors in each most recent field respective to a field just previous, and said statistical variables also including other statistical variables besides said isolativities and instabilities; field motion vector generation means for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, and generating a respective field motion vector by weighted averaging of the local motion vectors within each field, said weighted averaging including a multiplying of each of said local motion vectors within each field by a corresponding one of said isolation weights to form a product that selectively is included or is not included at all within a vector sum of weighted local motion vectors within each field, which vector sum is divided by a sum of the weights just of said weighted local motion vectors included in said vector sum in order to generate said respective field motion vector; accumulated motion vector generation means for receiving said field motion vector and generating said accumulated motion vector by accumulating said received field motion vector into an attenuated response to the previous accumulated motion vector supplied by attenuator means included within said accumulated motion vector generation means, and compensated accumulated motion vector generation means for receiving said accumulated motion vector, generating an error motion compensating value responsive to the magnitude of said accumulated motion vector generated by said accumulated motion vector generation means, and generating said compensated accumulated motion vector by coring said accumulated motion vector, so as to reduce said compensated accumulated motion vector to zero whenever the amplitude of said accumulated motion vector is smaller than said error motion compensating value--wherein said field motion vector generation means comprises; first comparing means for comparing said average correlation value with a reference value of low contrast, i.e., a first threshold value, and generating an abnormal-condition signal when said average correlation value is smaller than said first threshold value; second comparing means for generating a first difference signal by subtracting said average correlation value from said maximum correlation value, comparing said first difference signal with a reference value determining a moving object, i.e., a second threshold, and generating an abnormal-condition signal when said first difference signal is smaller than said second threshold value; third comparing means for generating a second difference signal by subtracting said second maximum correlation value from said maximum correlation value, comparing said second difference signal with a reference value determining a repeated image, i.e., a third threshold value, and generating an abnormal-condition signal when said second difference signal is smaller than said third threshold value; means for receiving the respective isolativity of each said local motion vector and generating an isolation weight signal that is relatively small when said isolativity increases and that is relatively large when said isolativity decreases, said isolation weight being generated in a predefined inverse relationship to said isolativity; means for receiving the respective instability of each said local motion vector and generating a stability weight signal that is relatively small when said instability increases and is relatively large when said instability decreases, said stability weight signal being generated in a predefined inverse relationship to said instability; and means for receiving said isolation weight signal and said stability weight signal of each said local motion vector, receiving an output of said reliability determination means as a control signal, and generating a respective output weight signal for each said local motion vector, which output weight signal is of zero value when said control signal is an abnormal-condition signal, and which output weight signal is generated by averaging said two received weight signals when said control signal is not an abnormal-condition signal--wherein said local motion vector generation means comprises; edge detection means for receiving said image data and converting field image into binary edge data by analyzing said received image data; storage means composed of at least two local motion estimation areas, for receiving said binary edge data and making one-field delay by sequentially extracting block data on reference areas of said local motion estimation areas; edge pattern matching means for receiving binary edge data of current field outputted from said edge detection means as compared data and binary edge data of previous field outputted from said storage means as reference data, matching patterns of said two block data, thereby sequentially generating correlation data of corresponding local motion estimation areas; and local motion vector estimation means for receiving said correlation data, and comparing said correlation data received from corresponding local motion estimation area with previous statistical variables detected in corresponding local motion estimation area, thereby generating a state of changed correlation data as statistical variables and at the same time, generating a position value of maximum correlation value as a local motion vector of corresponding local motion estimation area--and wherein said local motion vector estimation means comprises; comparing means for comparing said received correlation data with a previous maximum correlation value of corresponding local motion estimation area, thereby generating a first comparing signal when said correlation data is larger, local motion vector detection means for receiving address clocks and an output of said comparing means, generating a position value of said received correlation data by counting said address clocks, and generating a position value of said received correlation data as a local motion vector of corresponding local motion estimation area when said first comparing signal is received, and means for generating said statistical variables of corresponding local motion estimation area, which means comprises; maximum correlation generation means for receiving said received correlation data and an output of comparing means and generating said received correlation data as a maximum correlation value when said first comparing signal is received, means for receiving outputs of said maximum correlation generation means and said comparing means and generating a previous maximum correlation value as a second maximum correlation value when said first comparing signal is received, means for receiving said maximum correlation value and said second maximum correlation value and generating a maximum correlation difference by subtracting said second maximum correlation value from said maximum correlation value, and means for generating an average correlation value by adding said received correlation data into previous correlation data and dividing added correlation data by a number of motion vector candidates. - View Dependent Claims (17, 18, 19, 20, 21)
-
-
22. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field, and sequentially generating local motion vectors of said corresponding local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field and including the respective instability of each of said local motion vectors in each most recent field respective to a field just previous; field motion vector generation means for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, generating a respective stability weight signal that is of decreased value with increased instability, generating a respective output weight signal for each said local motion vector by additively combining its respective isolation and respective stability weights and generating a respective field motion vector by weighted averaging of the local motion vectors within each field, said weighted averaging including a multiplying of each of said local motion vectors within each field by a corresponding one of said output weight signals to form a product that selectively is included or is not included at all within a vector sum of weighted local motion vectors within each field, which vector sum is divided by a sum of the weights just of said weighted local motion vectors included in said vector sum in order to generate said respective field motion vector; compensated field motion vector generation means for receiving said field motion vector and a previous accumulated motion vector, generating an error motion compensating value responsive to the magnitude of said previous accumulated motion vector, and generating said compensated field motion vector by coring said field motion vector, so as to reduce said compensated field motion vector to zero whenever the amplitude of said field motion vector is smaller than said error motion compensating value; and accumulated motion vector generation means for receiving said compensated field motion vector and generating said accumulated motion vector by accumulating said received compensated field motion vector into said previous accumulated motion vector as attenuated by attenuator means included within said accumulated motion vector generation means. - View Dependent Claims (23)
-
-
24. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field, and sequentially generating local motion vectors of said corresponding local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity and the respective instability of each of said local motion vectors; field motion vector generation means for receiving and sequentially storing said local motion vectors and statistical variables generating a respective isolation weight for each of said stored local motion vectors inversely related to its respective isolativity over a range and zero-valued for larger isolativity values, generating a respective stability weight for each of said stored local motion vectors inversely related to its respective instability, and generating a respective field motion vector from a weighted average of said local motion vectors with each field, with each local motion vector being weighted by a result of additively mixing its respective isolation and stability weights; compensated field motion vector generation means for receiving said field motion vector and a previous accumulated motion vector, generating an error motion compensating value responsive to the magnitude of said previous accumulated motion vector, and generating said compensated field motion vector by coring said field motion vector, so as to reduce said compensated field motion vector to zero whenever the amplitude of said field motion vector is smaller than said error motion compensating value; and accumulated motion vector generation means for receiving said compensated field motion vector and generating said accumulated motion vector by accumulating said received compensated field motion vector into said previous accumulated motion vector as attenuated by attenuator means included within said accumulated motion vector generation means--wherein said local motion vector generation means comprises; edge detection means for receiving said image data and converting field image into binary edge data by analyzing said received image data; storage means composed of at least two local motion estimation areas, for receiving said binary edge data and making one-field delay by sequentially extracting block data on reference areas of said local motion estimation areas; edge pattern matching means for receiving binary edge data of current field outputted from said edge detection means as compared data and binary edge data of previous field outputted from said storage means as reference data, and matching patterns of said two block data, thereby sequentially generating correlation data of corresponding local motion estimation areas; and local motion vector estimation means for receiving said correlation data, and comparing said correlation data received from corresponding local motion estimation area with previous statistical variables detected in corresponding local motion estimation area, thereby generating a state of changed correlation data as statistical variables and at the same time, generating a position value of maximum correlation as a local motion vector of corresponding local motion estimation area--and wherein said local motion vector estimation means comprises; comparing means for comparing said received correlation data with a previous maximum correlation value of corresponding local motion estimation area, thereby generating a first comparing signal when said correlation data is larger; local motion vector detection means for receiving address clocks and an output of said comparing means, generating a position value of said received correlation data by counting said address clocks, and generating a position value of said received correlation data as a local motion vector of corresponding local motion estimation area when said first comparing signal is received; and means for generating said statistical variables of corresponding local motion estimation area, which means comprises; maximum correlation generation means for receiving said received correlation data and an output of comparing means and generating said received correlation data as a maximum correlation value when said first comparing signal is received; means for receiving outputs of said maximum correlation generation means and said comparing means and generating a previous maximum correlation value as a second maximum correlation value when said first comparing signal is received; means for receiving said maximum correlation value and said second maximum correlation value and generating a maximum correlation difference by subtracting said second maximum correlation value from said maximum correlation value; and means for generating an average correlation value by adding said received correlation data into previous correlation data and dividing added correlation data by a number of motion vector candidates. - View Dependent Claims (25, 26, 27, 28, 29, 30, 31)
-
-
32. An adaptive motion vector decision method for a digital image stabilization system having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising the steps of:
-
converting said received image data in fields into binary edge data; receiving said binary edge data, extracting reference block data of local motion estimation areas, and sequentially storing extracted reference block data in corresponding reference data storage areas; receiving said converted current-field binary edge data as compared block data, and sequentially generating correlation data in local motion estimation areas by matching patterns of said reference block data of said corresponding local motion estimation area; generating statistical variables and local motion vectors of corresponding local motion estimation areas by analyzing said sequentially received correlation data, said statistical variables including the isolativity of each of said local motion vectors from other local motion vectors for the same field, the stability of each of said local motion vectors respective to the correspondingly located local motion vector of the previous field, and other statistical variables; sequentially storing said received local motion vectors and statistical variables; sequentially checking said stored local motion vectors when a field search ending signal is received; determining from the stored corresponding statistical variables for each of said stored local motion vectors whether said image data in the local motion estimation area generating that said local motion vector was of irregular condition, as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector, or was of regular condition, as would cause that said local motion vector to be sufficiently reliable for consideration when generating a field motion vector; generating a weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition, but generating a zero-valued weight signal for each of said local motion vectors determined to be generated from image data of irregular condition, the step for generating said weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition being performed by additive rather than multiplicative processing; generating a respective field motion vector for each field period, by multiplying by corresponding weight signals said local motion vectors sequentially received within each field period, summing the resulting product vectors, and dividing the resulting sum vector by the sum of the respective weight signals of the local motion vectors sequentially received within that same field period; generating said accumulated motion vector, by substeps of; attenuating a previous accumulated motion vector, so as to result in its being multiplied by a predefined factor, and adding said attenuated previous accumulated motion vector and said field motion vector to obtain said accumulated motion vector. - View Dependent Claims (33, 34, 35, 36)
-
-
37. An adaptive motion vector decision method for a digital image stabilization system having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising the steps of:
-
converting said received image data in field into binary edge data; receiving said binary edge data, extracting reference block data of local motion estimation areas, and sequentially storing said extracted reference block data in corresponding reference data storage areas; receiving said converted current-field binary edge data as compared block data, and sequentially generating correlation data in local motion estimation areas by matching patterns of said reference block data of said corresponding local motion estimation area; generating statistical variables and local motion vectors of corresponding local motion estimation areas by analyzing said sequentially received correlation data, said statistical variables including the isolativity of each of said local motion vectors from other local motion vectors for the same field, the stability of each of said local motion vectors respective to the correspondingly located local motion vector of the previous field, and other statistical variables; sequentially storing said received local motion vectors and statistical variables; sequentially checking said stored local motion vectors when a field search ending signal is received; determining from the stored corresponding statistical variables for each of said stored local motion vectors whether said image data in the local motion estimation area generating that said local motion vector was of irregular condition, as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector, or was of regular condition, as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector; generating a weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition, but generating a zero-valued weight signal for each of said local motion vectors determined to be generated from image data of irregular condition, the step for generating said weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition being performed by additive rather than multiplicative processing; generating a respective field motion vector for each field period, by multiplying by corresponding weight signals said local motion vectors sequentially received within each field period, summing the resulting product vectors, and dividing the resulting stun vector by the sum of the respective weight signals of the local motion vectors sequentially received within that same field period; generating an error motion compensating value inversely related to the magnitude of a previous accumulated motion vector; generating a respective compensated field motion vector by coring each said field motion vector, so as to reduce said compensated field motion vector to zero whenever said field motion vector is smaller than said error motion compensating value; and generating said accumulated motion vector, by attenuating a previous accumulated motion vector by a predefined rate and adding said attenuated previous accumulated motion vector into said compensated field motion vector. - View Dependent Claims (38, 39, 40)
-
-
41. An adaptive motion vector decision method for a digital image stabilization system having a memory for storing image data in fields, and a control portion for connecting motion of image data stored in said memory by a received accumulated motion vector, comprising the steps of:
-
converting said received image data in fields into binary edge data; receiving said binary edge data, extracting reference block data of local motion estimation areas, and sequentially storing extracted reference block data in corresponding reference data storage areas; receiving said converted current-field binary edge data as compared block data and sequentially generating correlation data in local motion estimation areas by matching patterns of said reference block data of said corresponding local motion estimation area; generating statistical variables and local motion vectors of corresponding local motion estimation areas by analyzing said sequentially received correlation data said statistical variables including the isolativity of each of said local motion vectors from other local motion vectors for the same field, the stability of each of said local motion vectors respective to the correspondingly located local motion vector of the previous field, and other statistical variables; sequentially storing said received local motion vectors and statistical variables; sequentially checking said stored local motion vectors when a field search ending signal is received; determining from the stored corresponding statistical variables for each of said stored local motion vectors whether said image data in the local motion estimation area generating that said local motion vector was of irregular condition as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector, or was of regular condition, as would cause that said local motion vector to be sufficiently reliable for consideration when generating a field motion vector; generating a weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition, but generating a zero-valued weight signal for each of said local motion vectors determined to be generated from image data of irregular condition; generating a respective field motion vector for each field period, by multiplying by corresponding weight signals said local motion vectors sequentially received within each field period, summing the resulting product vectors, and dividing the resulting sum vector by the sum of the respective weight signals of the local motion vectors sequentially received within that same field period; generating said accumulated motion vector, by substeps of attenuating a previous accumulated motion vector, so as to result in its being multiplied by a predefined factor, and adding said attenuated previous accumulated motion vector and said field motion vector to obtain said accumulated motion vector; determining when said field motion vectors have been of substantially identical direction over at least a predetermined number of fields to generate a decision that there is deliberate camera movement in the most recent field; determining whether or not said received accumulated motion vector tends to leave a range in which undesired motion of said image data stored in said memory can be corrected for; selecting a first gain factor of about 0.995 for said step of attenuating said previous accumulated motion vector, unless it is decided that there is deliberate camera movement in the most recent field or it is determined that said received accumulated motion vector tends to leave a range in which undesired motion of said image data stored in said memory can be corrected for; and selecting a second gain factor of about 0.97 for said step of attenuating said previous accumulated motion vector when it is decided that there is deliberate camera movement in the most recent field or it is determined that said received accumulated motion vector tends to leave a range in which undesired motion of said image data stored in said memory can be corrected for.
-
-
42. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation circuitry receiving image data for predefined local motion estimation areas within each of successive fields and generating respective local motion vectors descriptive of field-to-field change of image data in said local motion estimation areas and respective statistical variables descriptive of said local motion vectors, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field determined by the magnitude of the result of differentially comparing that said local motion vector with the others of said local motion vectors for the same field period; comparator circuitry for receiving ones of said statistical variables descriptive of said local motion vectors, other than the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, and comparing said other statistical variables descriptive of each of said local motion vectors to threshold values for said other statistical variables for determining whether each of said local motion vectors is sufficiently reliable for consideration when generating a field motion vector or is insufficiently reliable for consideration when generating a field motion vector; field motion vector generator circuitry for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, supplying a respective output weight signal that is zero-valued for each local motion vector determined to be insufficiently reliable for consideration when generating a field motion vector, and otherwise supplying a respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector, which respective output weight signal depends on the respective isolation weight for that said local motion vector, and generating a field motion vector for each of said successive fields by said local motion vectors within said field in accordance with their respective output weights; and accumulation circuitry for generating a respective accumulated motion vector for each of said successive fields, said accumulation circuitry comprising an adder for generating a respective accumulated motion vector for each said successive field as the sum of the field motion vector with an attenuated response to the accumulated motion vector for the previous field and attenuator circuitry for generating said attenuated response to the accumulated motion vector for the previous field, wherein said local motion vector generation circuitry is of a type generating said statistical variables so as to include, in addition to the respective isolativity of each of said local motion vectors and said other said statistical variables, the respective instability of each of said local motion vectors, as determined by the magnitude of the result of differentially comparing that said local motion vector with the field motion vector for the preceding field period;
wherein said field motion vector generator circuitry is of a type generating a respective stability weight for each said local motion vector that is of decreased value with increased instability, and supplying a respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector, which respective output weight signal depends on a summation of the respective isolation weight and the respective stability weight for that said local motion vector, rather than on a product of the respective isolation weight and the respective stability weight for that said local motion vector. - View Dependent Claims (43, 44, 45, 46)
-
-
47. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation circuitry receiving image data for predefined local motion estimation areas within each of successive fields and generating respective local motion vectors descriptive of field-to-field change of image data in said local motion estimation areas and respective statistical variables descriptive of said local motion vectors, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field determined by the magnitude of the result of differentially comparing that said local motion vector with the others of said local motion vectors for the same field period; comparator circuitry for receiving ones of said statistical variables descriptive of said local motion vectors, other than the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, and comparing said other statistical variables descriptive of each of said local motion vectors to threshold values for said other statistical variables for determining whether each of said local motion vectors is sufficiently reliable for consideration when generating a field motion vector or is insufficiently reliable for consideration when generating a field motion vector; field motion vector generator circuitry for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, supplying a respective output weight signal that is zero-valued for each local motion vector determined to be insufficiently reliable for consideration when generating a field motion vector, and otherwise supplying a respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector, which respective output weight signal depends on the respective isolation weight for that said local motion vector, and generating a field motion vector for each of said successive fields by said local motion vectors within said field in accordance with their respective output weights; and
accumulation circuitry for generating a respective accumulated motion vector for each of said successive fields, said accumulation circuitry comprisingan adder for generating a respective accumulated motion vector for each said successive field as the sum of the field motion vector with an attenuated response to the accumulated motion vector for the previous field and attenuator circuitry for generating said attenuated response to the accumulated motion vector for the previous field which attenuator circuitry is of a controlled type, providing a less attenuated response responsive to a first condition of an attenuation control signal for said attenuator circuitry and providing a more attenuated response responsive to a second condition of said attenuation control signal for said attenuator circuitry; a threshold detector for detecting when said received accumulated motion vector supplied to said control portion of said image stabilization device has more than a prescribed magnitude to generate a threshold detector output signal of a second logic sense, said threshold detector output signal otherwise being of a first logic sense; and logic circuitry responding to said threshold detector output signal being of said second logic sense for generating said second condition of said attenuation control signal and at least conditionally responding to said threshold detector output signal being of said first logic sense for generating said first condition of said attenuation control signal. - View Dependent Claims (48, 49, 50, 51, 52)
-
-
53. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation circuitry receiving image data for predefined local motion estimation areas within each of successive fields and generating respective local motion vectors descriptive of field-to-field change of image data in said local motion estimation areas and respective statistical variables descriptive of said local motion vectors, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field determined by the magnitude of the result of differentially comparing that said local motion vector with the others of said local motion vectors for the same field period; comparator circuitry for receiving ones of said statistical variables descriptive of said local motion vectors, other than the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, and comparing said other statistical variables descriptive of each of said local motion vectors to threshold values for said other statistical variables for determining whether each of said local motion vectors is sufficiently reliable for consideration when generating a field motion vector or is insufficiently reliable for consideration when generating a field motion vector; field motion vector generator circuitry for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, supplying a respective output weight signal that is zero-valued for each local motion vector determined to be insufficiently reliable for consideration when generating a field motion vector, and otherwise supplying a respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector, which respective output weight signal depends on the respective isolation weight for that said local motion vector, and generating a field motion vector for each of said successive fields by said local motion vectors within said field in accordance with their respective output weights; and accumulation circuitry for generating a respective accumulated motion vector for each of said successive fields, said accumulation circuitry comprising an adder for generating a respective accumulated motion vector for each said successive field as the sum of the field motion vector with an attenuated response to the accumulated motion vector for the previous field and attenuator circuitry for generating said attenuated response to the accumulated motion vector for the previous field; and compensated accumulated motion vector generation circuitry, which said compensated accumulated motion vector generation circuitry generates a respective compensated accumulated motion vector in controlled cored response to each said accumulated motion vector and comprises; circuitry for generating a first error motion compensating value that decreases from a prescribed value dependent on increase in the magnitude of said accumulated motion vector when the magnitude of said accumulated motion vector is within a range between zero and a prescribed value at and beyond which said first error motion compensating value is zero-valued; circuitry for generating a second error motion compensating value that decreases from a prescribed value dependent on increase in the magnitude of said average maximum correlation difference, when the magnitude of said average maximum correlation difference is within a range between zero and a prescribed value at and beyond which said second error motion compensating value is zero-valued; averaging circuitry averaging said first error motion compensating value and said second error motion compensating value for generating an averaged motion compensating value; and controlled coring circuitry supplying said respective compensated accumulated motion vector in cored response to said accumulated motion vector supplied from said accumulation circuitry for generating a respective accumulated motion vector for each of said successive fields, which cored response is zero-valued except when the magnitude of said accumulated motion vector supplied from said accumulation circuitry exceeds said averaged motion compensating value. - View Dependent Claims (54)
-
-
55. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation circuitry receiving image data for predefined local motion estimation areas within each of successive fields and generating respective local motion vectors descriptive of field-to-field change of image data in said local motion estimation areas and respective statistical variables descriptive of said local motion vectors, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field determined by the magnitude of the result of differentially comparing that said local motion vector with the others of said local motion vectors for the same field period; comparator circuitry for receiving ones of said statistical variables descriptive of said local motion vectors, other than the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, and comparing said other statistical variables descriptive of each of said local motion vectors to threshold values for said other statistical variables for determining whether each of said local motion vectors is sufficiently reliable for consideration when generating a field motion vector or is insufficiently reliable for consideration when generating a field motion vector; field motion vector generator circuitry for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, supplying a respective output weight signal that is zero-valued for each local motion vector determined to be insufficiently reliable for consideration when generating a field motion vector, and otherwise supplying a respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector, which respective output weight signal depends on the respective isolation weight for that said local motion vector, and generating a field motion vector for each of said successive fields by said local motion vectors within said field in accordance with their respective output weights, compensated field motion vector generation circuitry, which said compensated field motion vector generation circuitry generates a respective compensated field motion vector in controlled cored response to each said field motion vector supplied by said field motion vector generator circuitry; and accumulation circuitry for generating a respective accumulated motion vector for each of said successive fields, said accumulation circuitry comprising an adder for generating a respective accumulated motion vector for each said successive field as the sum of the compensated field motion vector with an attenuated response to the accumulated motion vector for the previous field and attenuator circuitry for generating said attenuated response to the accumulated motion vector for the previous field;
wherein said local motion vector generation circuitry is of a type generating said statistical variables so as to include, in addition to the respective isolativity of each of said local motion vectors and said other said statistical variables, the respective instability of each of said local motion vectors, as determined by the magnitude of the result of differentially comparing that said local motion vector with the field motion vector for the preceding field period;
wherein said field motion vector generator circuitry is of a type generating a respective stability weight for each said local motion vector that is of decreased value with increased instability and supplying, a respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector; and
wherein the respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector, depends on a summation of the respective isolation weight and the respective stability weight for that said local motion vector, rather than on a product of the respective isolation weight and the respective stability weight for that said local motion vector. - View Dependent Claims (56, 57, 58, 59, 60)
-
-
61. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, said image stabilization device comprising:
-
local motion vector generation circuitry receiving image data for predefined local motion estimation areas within each of successive fields and generating respective local motion vectors descriptive of field-to-field change of image data in said local motion estimation areas and respective statistical variables descriptive of said local motion vectors, said statistical variables including the respective isolativity of each of said local motion vectors from other local motion vectors within the same field determined by the magnitude of the result of differentially comparing that said local motion vector with the others of said local motion vectors for the same field period; comparator circuitry for receiving ones of said statistical variables descriptive of said local motion vectors, other than the respective isolativity of each of said local motion vectors from other local motion vectors within the same field, and comparing said other statistical variables descriptive of each of said local motion vectors to threshold values for said other statistical variables for determining whether each of said local motion vectors is sufficiently reliable for consideration when generating a field motion vector or is insufficiently reliable for consideration when venerating a field motion vector; field motion vector generator circuitry for receiving said local motion vectors and statistical variables, generating a respective isolation weight for each said local motion vector that is of decreased value with increased isolativity and is zero for isolativity beyond a predefined value, supplying a respective output weight signal that is zero-valued for each local motion vector determined to be insufficiently reliable for consideration when generating a field motion vector, and otherwise supplying a respective output weight signal generated for each local motion vector determined to be sufficiently reliable for consideration when generating a field motion vector, which respective output weight signal depends on the respective isolation weight for that said local motion vector, and generating a field motion vector for each of said successive fields by said local motion vectors within said field in accordance with their respective output weights; compensated field motion vector generation circuitry, which said compensated field motion vector generation circuitry generates a respective compensated field motion vector in controlled cored response to each said field motion vector supplied by said field motion vector generator circuitry; and accumulation circuitry for generating a respective accumulated motion vector for each of said successive fields, said accumulation circuitry comprising an adder for generating a respective accumulated motion vector for each said successive field as the sum of the compensated field motion vector with an attenuated response to the accumulated motion vector for the previous field and attenuator circuitry for generating said attenuated response to the accumulated motion vector for the previous field, wherein said attenuator circuitry for generating said attenuated response to the accumulated motion vector for the previous field is of a controlled type, providing a less attenuated response responsive to a first condition of an attenuation control signal for said attenuator circuitry and providing a more attenuated response responsive to a second condition of said attenuation control signal for said attenuator circuitry; a threshold detector for detecting when said received accumulated motion vector supplied to said control portion of said image stabilization device has more than a prescribed magnitude to generate a threshold detector output signal of a second logic sense, said threshold detector output signal otherwise being of a first logic sense; and logic circuitry responding to said threshold detector output signal being of said second logic sense for generating said second condition of said attenuation control signal and at least conditionally responding to said threshold detector output signal being of said first logic sense for generating said first condition of said attenuation control signal. - View Dependent Claims (62, 63, 64, 65, 66)
-
-
67. An adaptive motion vector decision method for a digital image stabilization system having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising the steps of:
-
determining, from said received image data as fall in each field into local motion estimation areas, respective local motion vectors and statistical variables descriptive of said local motion vectors; determining a respective field motion vector for each field from said local motion vectors and statistical variables descriptive of said local motion vectors; determining when field motion vectors have been of substantially identical direction over at least a predetermined number of fields; determining when the magnitude of said received accumulated motion vector exceeds a prescribed threshold value; and generating said accumulated motion vector, by substeps of; selecting a first gain factor of about 0.995 except when said received accumulated motion vector exceeds said prescribed threshold value or field motion vectors have been of substantially identical direction over at least a predetermined number of fields, selecting a second gain factor of about 0.97 when said received accumulated motion vector exceeds said prescribed threshold value or said field motion vectors have been of substantially identical direction over at least said predetermined number of fields, attenuating a previous accumulated motion vector, so as to result in its being multiplied by the selected one of said first and second gain factors, and adding said attenuated previous accumulated motion vector and said field motion vector to obtain said accumulated motion vector.
-
-
68. An adaptive motion vector decision method for a digital image stabilization system having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising the steps of:
-
determining, from said received image data as fall in each field into local motion estimation areas, respective local motion vectors and statistical variables descriptive of said local motion vectors; determining from the corresponding statistical variables for each of said local motion vectors whether said image data in the local motion estimation area generating that said local motion vector was of irregular condition, as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector, or was of regular condition, as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector; generating a weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition, but generating a zero-valued weight signal for each of said local motion vectors determined to be generated from image data of irregular condition the step for venerating said weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition being performed by additive rather than multiplicative processing; performing a weighted averaging of said local motion vectors within each field for generating a respective field motion vector; and generating said accumulated motion vector, by substeps of; attenuating a previous accumulated motion vector, so as to result in its being multiplied by a predefined factor, and adding said attenuated previous accumulated motion vector and said field motion vector to obtain said accumulated motion vector.
-
-
69. An adaptive motion vector decision method for a digital image stabilization system having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising the steps of:
-
determining, from said received image data as fall in each field into local motion estimation areas, respective local motion vectors and statistical variables descriptive of said local motion vectors; determining from the corresponding statistical variables for each of said local motion vectors whether said image data in the local motion estimation area generating that said local motion vector was of irregular condition, as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector, or was of regular condition, as would cause that said local motion vector to be insufficiently reliable for consideration when generating a field motion vector; generating a weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition, but generating a zero-valued weight signal for each of said local motion vectors determined to be generated from image data of irregular condition the step for generating said weight signal dependent on isolativity and stability for each of said local motion vectors determined to be generated from image data of regular condition being performed by additive rather than multiplicative processing; performing a weighted averaging of said local motion vectors within each field for generating a respective field motion vector; coring said field motion vector; and generating said accumulated motion vector, by substeps of; attenuating a previous accumulated motion vector, so as to result in its being multiplied by a predefined factor, and adding said attenuated previous accumulated motion vector and said cored field motion vector to obtain said accumulated motion vector.
-
-
70. An image stabilization device having a memory for storing image data in fields, and a control portion for correcting motion of image data stored in said memory by a received accumulated motion vector, comprising:
-
local motion vector generation means for receiving image data, converting said received image data into binary edge data, sequentially generating correlation data in local motion estimation areas by matching patterns of said edge data of current field and previous field, and sequentially generating local motion vectors of said corresponding local motion estimation areas and respective statistical variables descriptive of said local motion vectors by analyzing said generated correlation data, said statistical variables including the respective isolativity and the respective instability of each of said local motion vectors; field motion vector generation means for receiving and sequentially storing said local motion vectors and statistical variables, generating a respective isolation weight for each of said stored local motion vectors inversely related to its respective isolativity over a range and zero-valued for larger isolativity values, generating a respective stability weight for each of said stored local motion vectors inversely related to its respective instability, and generating a respective field motion vector from a weighted average of said local motion vectors with each field, with each local motion vector being weighted by a result of additively mixing its respective isolation and stability weights; compensated field motion vector generation means for receiving said field motion vector and a previous accumulated motion vector, generating an error motion compensating value responsive to the magnitude of said previous accumulated motion vector, and generating said compensated field motion vector by coring said field motion vector, so as to reduce said compensated field motion vector to zero whenever the amplitude of said field motion vector is smaller than said error motion compensating value; and accumulated motion vector generation means for receiving said compensated field motion vector and generating said accumulated motion vector by accumulating said received compensated field motion vector into said previous accumulated motion vector as attenuated by attenuator means included within said accumulated motion vector generation means--wherein said local motion vector generation means comprises; edge detection means for receiving said image data and converting field image into binary edge data by analyzing said received image data; storage means for temporarily storing said binary edge data for at least selected pixels in each of at least two local motion estimation areas, for receiving said binary edge data for each of said local motion estimation areas and storing that data until at least a next succeeding field; edge pattern matching means for receiving binary edge data of current field supplied from said edge detection means as compared data and binary edge data of a previous field supplied from said storage means as reference data, and matching patterns of said two block data, thereby sequentially generating correlation data of corresponding local motion estimation areas; and local motion vector estimation means for receiving said correlation data, and comparing said correlation data received from corresponding local motion estimation area with previous statistical variables detected in corresponding local motion estimation area, thereby generating a state of changed correlation data as statistical variables and at the same time, generating a position value of maximum correlation as a local motion vector of corresponding local motion estimation area--and wherein said local motion vector estimation means comprises; comparing means for comparing said received correlation data with a previous maximum correlation value of corresponding local motion estimation area, thereby generating a first comparing signal when said correlation data is larger; local motion vector detection means for receiving address clocks and an output of said comparing means, generating a position value of said received correlation data by counting said address clocks, and generating a position value of said received correlation data as a local motion vector of corresponding local motion estimation area when said first comparing signal is received; and means for generating said statistical variables of corresponding local motion estimation area, which means comprises; maximum correlation generation means for receiving said received correlation data and an output of comparing means and generating said received correlation data as a maximum correlation value when said first comparing signal is received; means for receiving outputs of said maximum correlation generation means and said comparing means and generating a previous maximum correlation value as a second maximum correlation value when said first comparing signal is received; means for receiving said maximum correlation value and said second maximum correlation value and generating a maximum correlation difference by subtracting said second maximum correlation value from said maximum correlation value; and means for generating an average correlation value by adding said received correlation data into previous correlation data and dividing added correlation data by a number of motion vector candidates.
-
Specification