Motion descriptor generating apparatus by using accumulated motion histogram and a method therefor

• US 6,597,738 B1
• Filed: 08/27/1999
• Issued: 07/22/2003
• Est. Priority Date: 02/01/1999
• Status: Expired due to Term

• Current Assignee: Pantech Incorporated
• Original Assignee: Hyundai Curitel ,inc
• Inventors: Moon, Joo-Hee, Kim, Hae-Kwang, Park, Cheol-Soo
• Primary Examiner(s): Philippe, Gims S.
• Application Number:
  US09/384,163
• Time in Patent Office: 1,425 Days
• Field of Search: 375/240.09, 375/240.16, 375/240.24, 375/240.08, 348/420.1, 348/422.1, 348/402.1, 348/413.1, 348/416.1, 348/431.1, 348/699, 348/700, 382/168, 382/107, 382/225, 382/170, 382/173, 382/305, 382/236, 382/243, 345/428
• US Class Current: 375/240.16
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 7/223 using block-matching

Abstract

A motion descriptor generating apparatus using accumulated motion histogram and a method therefor. Accumulated motion histograms are generated and motion descriptors are generated by using the accumulated motion histograms. The motion descriptor generating apparatus uses an accumulated motion histogram and includes a motion histogram generating unit for respectively generating a motion histogram with relation to intensity data and direction data of an input motion. An accumulated motion histogram generating unit generates a two-dimensional accumulated motion histogram in a predetermined sequence by using the motion histogram which is generated in the motion histogram generating unit. A motion descriptor generating unit structures (hierarchy structure) video into certain units according to a change amount of the accumulated motion histogram based on a lapse of time, which is generated in the accumulated motion intensity histogram generating unit. A motion descriptor is generated which describes motion characteristics with relation to the respective structured units.

Claims

1. A motion descriptor generating apparatus by using accumulated motion histogram, comprising:

motion histogram generating means for respectively generating motion histogram with relation to intensity data and direction data of an input motion in video;

accumulated motion histogram generating means for accumulating the motion histogram and generating accumulated motion histogram in a predetermined sequence; and

motion descriptor generating means for structuring hierarchy of the video by classifying a unit of the video according to a change amount of the accumulated motion histogram for a predetermined time, which is generated in said accumulated motion intensity histogram generating means, and generating a motion descriptor which describes motion characteristics of the unit of the video based on the hierarchy of the video,

wherein said motion descriptor comprises a video ID descriptor, a motion descriptor for describing temporal position of video which is described by a corresponding motion descriptor, a motion description level for displaying a characteristic information describing level of the motion which is described, a direction descriptor for describing three-dimensional characteristics information for the motion direction, a intensity descriptor for describing three-dimensional characteristics information for the motion intensity, selection flag information for direction and intensity which are usable in video search, an additional descriptor for describing motion characteristics of next stage, and flag and numbers of the additional descriptor.

2. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, wherein said motion descriptor generating means comprises:

a part for generating a clip motion descriptor using a clip which is indexed using the accumulated motion histogram of the accumulated motion histogram generating means; and

a part for generating motion descriptor by using the change amount or the motion histogram which is computed in said part for computing a change amount of accumulated motion histogram.

3. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 2, wherein said part for generating a clip motion descriptor comprising:

a part for computing a change amount of accumulated motion histogram which is generated in said accumulated motion histogram generating means

a clip-time index part for indexing temporal position and clip numbers of video which is computed in said part for computing change amount of motion histogram; and

a comparison part for comparing the change amount which is computed by said part for computing change amount of motion histogram with a predetermined threshold value, and enabling said part for computing the change amount of the motion histogram or said clip-time index.

4. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 2 or claim 3, wherein the motion descriptor, which is generated by said motion descriptor generating means, describes information onto time of total accumulated motion histogram data.

5. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 2, wherein the motion descriptor, which is generated by said motion descriptor generating means, describes number of bin of the accumulated motion histogram with relation to motion direction.

6. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 2, wherein the motion descriptor, which is generated by said motion descriptor generating means, describes number of clip.

7. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 2, wherein the motion descriptor, which is generated by said motion descriptor generating means, describes an average value of the total accumulated histogram with relation to the motion direction described by the clip motion descriptor.

8. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 2, wherein the motion descriptor, which is generated by said motion descriptor generating means, describes an average value of the motion intensity for each bin of the accumulated motion histogram with relation to the motion direction according to the described time.

17. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, further comprising:

divergence processing means of motion estimation value divergence for providing motion search value which is processed according to external frame selection mode to said motion histogram generating means.

18. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, further comprising:

motion filtering means for providing an input motion estimation value, of which intensity and direction are computed, to said motion histogram generating means.

19. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, further comprising:

motion estimation value divergence processing means for processing to generate motion estimation value according to external frame selection mode; and

motion filtering means for computing intensity and direction of the motion estimation values, which are processed in said motion estimation value divergence processing means, and filtering the motion estimation values to provide to said motion histogram generating means.

20. A motion descriptor generating apparatus by using accumulated motion histogram according to claim 17 or claim 19, wherein said divergence processing means of motion estimation value divergence comprises:

an average computation part for computing respective averages between a present first area which has a motion estimation value in a current input image, and a present second area to a present fourth area which are adjacent to and neighboring the present first area, or between the present first area, a previous first area and previous second to previous fourth areas neighboring the previous first area;

an absolute value computation part for respectively computing differences between the present first area and the present second to fourth areas, or between the present first area, and the previous first area and the previous second to fourth areas, and computing respective absolute values of the obtained differences; and

a motion vector values comparison/conversion part for comparing the absolute values which are respectively computed in said absolute value computation part, and performing conversion of X, Y motion vectors which are input according to the comparison result.

21. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 20, wherein said average value computation part computes respective average values between the present first area which has the motion estimation value in the current input image, and the present second to the fourth areas neighboring the present first area, when the frame motion mode is present image selection mode.

22. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 20, wherein said average value computation part computes respective average values between the present first area, and the previous first area and the previous second to fourth areas, when the frame motion mode is previous image selection mode.

23. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 20, wherein said motion vector values comparison/conversion part compares the absolute values of the computed differences between the present first area and the present second area and between the present first area and the present third area, and converting the input X, Y motion vector values into 0 if all the absolute values are smaller than a predetermined threshold value, when the frame motion mode is present image selection mode.

24. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 23, wherein said motion vector values comparison/conversion part outputs Y without any conversion and 0 after converting the input X motion vector values into 0, if the absolute value of the computed difference between the present first area and the present second area and the absolute value of the computed difference between the present first area and the present third area are not commonly smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the present third area is smaller than the predetermined threshold value.

25. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 23, wherein said motion vector values comparison/conversion part outputs X motion vector value without any conversion and 0 after converting the input Y motion vector value into 0, if the absolute value of the computed difference between the present first area and the present second area and the absolute value of the computed difference between the present first area and the present third area are not commonly smaller than the predetermined threshold value, if the absolute value of the difference between the present first area and the present third area is not smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the present second area is smaller than the predetermined threshold value.

26. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 23, wherein said motion vector values comparison/conversion part outputs X, Y motion vector values without any conversion, if the absolute value of the computed difference between the present first area and the present second area and the absolute value of the computed difference between the present first area and the present third area are not smaller than the predetermined threshold value, if the absolute value of the computed difference between the present first area and the present third area is not smaller than the predetermined threshold value, or if the absolute value of the computed difference between the present first area and the present second area is smaller than the predetermined threshold value.

27. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 20, wherein said motion vector values comparison/conversion part compares the absolute value of the computed difference between the present first area and the previous first area, and converting the respective input X, Y motion vector values into 0 if the absolute value is smaller than a predetermined threshold value, when the frame motion mode is previous image selection mode.

28. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 27, wherein said motion vector values comparison/conversion part outputs X, Y motion vector values after respectively converting them into 0, if the absolute value of the computed difference between the present first area and the previous first area is not smaller than the predetermined threshold value, or if all the absolute values of the computed differences between the present first area and the previous second area and between the present first area and the previous third area are commonly smaller than the predetermined threshold value.

29. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 27, wherein said motion vector values comparison/conversion part outputs Y motion vector value without any conversion and 0 after converting the input X motion vector value into 0 if the absolute value of the computed difference between the present first area and the previous first area, if the absolute values of the computed differences between the present first area and the previous second area and between the present first area and the previous third area are not smaller than the predetermined threshold value, or if the absolute value of the computed difference between the present first area and the previous third area is smaller than the predetermined threshold value.

30. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 20, wherein said motion vector values comparison/conversion part outputs X motion vector value without any conversion and 0 after converting the input Y motion vector value into 0, if the absolute value of the computed difference between the present first area and previous first area is not smaller than the predetermined threshold value, if all the absolute values of the computed differences between the present first area and the present second area and between the present first area and the present third area are not smaller than the predetermined threshold value, if the absolute value of the computed difference between the present first area and the present third area is not smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the previous second area is smaller than the predetermined threshold value.

31. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 20, wherein said motion vector values comparison/conversion part outputs X, Y motion vector values without any conversion, if the absolute value of the computed difference between the present first area and previous first area is not smaller than the predetermined threshold value, if all the absolute values of the computed differences between the present first area and the previous second area and between the present first area and the previous third area are not smaller than the predetermined threshold value, if the absolute value of the computed difference between the present first area and the previous third area is not smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the present second area is smaller than the predetermined threshold value.

32. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 18 or claim 19, wherein said motion filtering means comprises:

a motion intensity computation part for computing motion intensity by using the X,Y motion vector values which are processed by said motion estimation value divergence means;

a motion vector value comparison/conversion part for comparing motion intensity which is computed by said motion intensity computation part with a predetermined threshold value, and performing conversion of the motion vectors according to the comparison result;

a motion direction computation part for computing motion direction by using the motion vector values which are converted by said motion vector value comparison/conversion part; and

a motion direction quantization/inverse-quantization part for quantizing/inverse-quantizing motion direction which is computed by said motion direction computation part.

33. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 32, wherein said motion vector values comparison/conversion part converts X,Y motion vector values, which are outputted from said motion estimation value divergence part, into 0 when the motion intensity which is computed by said motion intensity computation part is smaller than a predetermined value, and outputs X,Y motion vector values, which are outputted from said motion estimation value divergence part, without any conversion when the motion intensity which is computed by said motion intensity computation part is larger than the predetermined value.

34. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 32, wherein said motion direction quantization/inverse-quantization part performs quantization of the motion direction which is computed by said motion direction computation part, by using either a linear quantization method or a non-linear quantization method.

53. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, wherein the video ID descriptor is to describe a credit for identifying video name, video producer, source and offerer, production date of the video and version of the video.

54. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, wherein the time descriptor includes a sequential time descriptor and a random time descriptor according to whether the motion descriptor represents all motion data in the description scope or not.

55. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 54, wherein the sequential time descriptor is to describe starting time, ending time, and duration of a motion data.

56. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 54, wherein the random time descriptor is to describe partial motion data and characteristics in the time descriptor.

57. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, wherein the motion direction descriptor is to describe a average of motion direction (θ

1), temporal distribution of motion direction (θ

, p), deviation average of motion direction (σ

θ

1), spatial distribution of motion direction (σ

θ

p), dominant motion direction, accumulated histogram of motion direction, and motion intensity average for motion direction of each bin.

58. A motion descriptor generating apparatus by using accumulated motion histogram according to claim 57, wherein the motion direction average (θ

1) represents an average of motion direction averages (ρ

) of T images in a scope which is intended by the time descriptor in the motion descriptor, which is obtained by below formula (10);

$\begin{array}{cc}{\mathrm{\theta <br><br>}}_1=\underset{i=0}{\stackrel{T}{\sum <br><br>}}<br><br>{\mathrm{\rho <br><br>}}_i/T& \left(10\right)\end{array}$

wherein T may not equal to numbers of all images intended by the time descriptor.

59. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 57, wherein the temporal distribution of the motion direction represents characteristics of temporal distribution and distortion degree with relation to the direction averages (ρ

) of respective images in the direction average (θ

1), which is obtained by below formula (11);

$\begin{array}{cc}{\mathrm{\theta <br><br>}}_p=\sqrt{\underset{i=0}{\stackrel{T}{\sum <br><br>}}<br><br>{\left({\mathrm{\rho <br><br>}}_i-\mathrm{\theta <br><br>}\right)}^p}/T,\text{\hspace{1em}<br><br>}<br><br>P\ge <br><br>2.& \left(11\right)\end{array}$

60. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 57, wherein the deviation average of the motion direction represents an average of the spatial distribution of directions with relation to total T images in the time, which is obtained by below formula (13):

$\begin{array}{cc}{\mathrm{\sigma <br><br>}}_{\mathrm{\theta <br><br>}<br><br>1}=\underset{i=0}{\stackrel{T}{\sum <br><br>}}<br><br>{\mathrm{\sigma <br><br>}}_{\mathrm{\tau <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>i}/T.& \left(13\right)\end{array}$

61. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 57, wherein the spatial distribution of motion direction represents characteristics of spatial distribution and distortion with relation to standard deviations (σ

τ

i) of respective images in standard deviation averages (σ

θ

1), which is obtained by below formula (14);

$\begin{array}{cc}{\mathrm{\sigma <br><br>}}_{\mathrm{\theta <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>p}=\sqrt{\underset{i=0}{\stackrel{T}{\sum <br><br>}}<br><br>\left({\mathrm{\sigma <br><br>}}_{\mathrm{\tau <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>i}-{\mathrm{\sigma <br><br>}}_{\mathrm{\theta <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>1}\right)}/T,\text{\hspace{1em}<br><br>}<br><br>P\ge <br><br>2.& \left(14\right)\end{array}$

62. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 57, wherein the dominant direction of motion represents a direction of which histogram value is largest among motion histograms which are generated with relation to direction by using below formula (5), wherein a plurality of dominant directions may be determined in sequence of the histogram value:

63. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 1, wherein the intensity descriptor is to describe motion intensity average (ω

1), temporal distribution of motion intensity (ω

q), deviation average of motion intensity (σ

ω

1), spatial distribution of motion intensity (σ

ω

q), accumulated histogram of motion intensity (Hω

), and average of motion intensity with relation to motion direction.

64. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 63, wherein the motion intensity average (ω

1) represents an average of motion intensity averages (λ

) of T images in a scope which is intended by the time descriptor in the motion descriptor, which is obtained by below formula (17);

$\begin{array}{cc}{\mathrm{\omega <br><br>}}_1=\left(\underset{i=0}{\stackrel{T}{\sum <br><br>}}<br><br>{\mathrm{\lambda <br><br>}}_i\right)/T& \left(17\right)\end{array}$

wherein T may not equal to numbers of all images in the description time.

65. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 63, wherein the temporal distribution of motion intensity (ω

q) is obtained by below formula (18);

$\begin{array}{cc}{\mathrm{\omega <br><br>}}_q=\sqrt{\underset{i=0}{\stackrel{T}{\sum <br><br>}}<br><br>{\left({\mathrm{\lambda <br><br>}}_i-{\mathrm{\omega <br><br>}}_1\right)}^q}/T,\text{\hspace{1em}<br><br>}<br><br>q\ge <br><br>2.& \left(18\right)\end{array}$

66. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 63, wherein the deviation average of the motion direction (σ

ω

1) represents an average of the spatial distribution of intensity with relation to total T images in the time, which is obtained by below formula (20);

$\begin{array}{cc}{\mathrm{\sigma <br><br>}}_{\mathrm{\omega <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>1}=\left(\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{\sigma <br><br>}}_{\mathrm{\lambda <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>i}\right)/T.& \left(20\right)\end{array}$

67. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 63, wherein the spatial distribution of motion intensity (σ

ω

q) is obtained by below formula (21);

$\begin{array}{cc}{\mathrm{\sigma <br><br>}}_{\mathrm{\omega <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>q}=\sqrt{\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\left({\mathrm{\sigma <br><br>}}_{\mathrm{\lambda <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>i}-{\mathrm{\sigma <br><br>}}_{\mathrm{\omega <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>1}\right)}^q}/T,\text{\hspace{1em}<br><br>}<br><br>q\ge <br><br>2.& \left(21\right)\end{array}$

68. The motion descriptor generating apparatus by using accumulated motion histogram according to claim 57 or claim 63, wherein the motion histogram descriptor, which describes accumulated histogram of motion direction and accumulated histogram of motion intensity, represents bin number of histogram, histogram value and representative value of bin.

84. The method for generating motion descriptors by using accumulated motion histogram according to claim 63, wherein the motion histogram descriptor, which describes accumulated histogram of motion direction and accumulated histogram of motion intensity, represents bin number of histogram, histogram value and representative value of bin.

9. A method for generating motion descriptors by using accumulated motion histogram, comprising the steps of:

generating motion histogram for respectively generating motion histogram with relation to intensity and direction data of an input motion in video;

generating accumulated motion histogram for accumulating the motion histogram and generating accumulated motion histogram; and

structuring hierarchy of the video by classifying a unit of the video according to a change amount of the accumulated motion histogram for a predetermined time, and generating a motion descriptor which describes motion characteristics of the unit of the video based on the hierarchy of the video,

wherein said motion descriptor comprises a video ID descriptor, a motion descriptor for describing temporal position of video which is described by a corresponding motion descriptor, a motion description level for displaying a characteristic information describing level of the motion which is described, a direction descriptor for describing three-dimensional characteristics information for the motion direction, a intensity descriptor for describing three-dimensional characteristics information for the motion intensity, selection flag information for direction and intensity which are usable in video search, an additional descriptor for describing motion characteristics of next stage, and flag and numbers of the additional descriptor.

10. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, the motion descriptor generating step comprises the steps of:

generating a clip motion descriptor using a clip which is indexed using the accumulated motion histogram of the accumulated motion histogram generating means; and

generating motion descriptor by using the change amount or the motion histogram which is computed in said part for computing a change amount of accumulated motion histogram.

11. The method for generating motion descriptors by using accumulated motion histogram according to claim 10, wherein said step of generating a clip motion descriptor comprising:

computing a change amount of accumulated motion histogram which is generated in the accumulated motion histogram generating step

clip-time indexing for generating clip motion descriptor by indexing temporal position and clip numbers of video which is computed in the computation step of motion histogram change amount; and

comparing the change amount which is computed by the computation step of motion histogram change amount with a predetermined threshold value, and repeating the computation step of the motion histogram change amount or the clip-time index step.

12. The method for generating motion descriptors by using accumulated motion histogram according to claim 10 or claim 11, wherein the motion descriptor, which is generated by said motion descriptor generation step, describes motion information onto temporal position of the video.

13. The method for generating motion descriptors by using accumulated motion histogram according to claim 10, wherein the motion descriptor, which is generated by the motion descriptor generation step, describes number of bin of the accumulated motion histogram with relation to motion direction.

14. The method for generating motion descriptors by using accumulated motion histogram according to claim 10, wherein the motion descriptor, which is generated by the motion descriptor generating step, describes number of clip.

15. The method for generating motion descriptors by using accumulated motion histogram according to claim 10, wherein the motion descriptor, which is generated by the motion descriptor generating step, describes an average value of the total accumulated histogram with relation to the motion direction described by the clip motion descriptor.

16. A method for generating motion descriptors by using accumulated motion histogram according to claim 10, wherein the motion descriptor, which is generated by the motion descriptor generating step, describes an average value of the motion intensity for each bin of the accumulated motion histogram with relation to the motion direction according to the described time.

35. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, further comprising a step of processing motion estimation value divergence for providing motion estimation value to said motion histogram generating means after processing it according to external frame selection mode.

36. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, further comprising a step of motion filtering for providing an input motion estimation value, of which intensity and direction are computed, to said step for generating motion histograms after filtering.

37. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, further comprising the steps of:

motion estimation value diverging for providing motion estimation value to said motion histogram generating means after processing it according to external frame selection mode; and

motion filtering for computing intensity and direction of the motion estimation values, which are processed in said motion estimation value divergence means, and filtering the motion estimation values to provide to said step of motion histogram generating.

38. The method for generating motion descriptors by using accumulated motion histogram according to claim 35 or claim 37, wherein said step of motion. estimation value diverging comprises the sub-steps of:

average computing for computing respective averages between a present first area which has a motion estimation value in a current input image, and a present second area to a present fourth area which are adjacent to and neighboring the present first area, or between the present first area, a previous first area and previous second to previous fourth areas neighboring the previous first area;

absolute value computing for respectively computing differences between the present first area and the present second to fourth areas, or between the present first area and the previous first area and between the present first area and the previous second to fourth areas, and computing respective absolute values of the obtained differences; and

motion vector values comparing/converting for comparing the absolute values which are respectively computed in said step of absolute value computing and performing conversion of X, Y motion vectors which are input according to the comparison result.

39. The method for generating motion descriptors by using accumulated motion histogram according to claim 38, wherein said step of average value computing comprises computing respective average values between the present first area which has the motion estimation value in the current input image, and the present second to the fourth areas neighboring the present first area, when the frame motion mode is present image selection mode.

40. The method for generating motion descriptors by using accumulated motion histogram according to claim 38, wherein said step of average value computing comprises computing respective average values between the present first area and the previous first area and between the present first area and the previous second to fourth areas which are neighboring the previous first area, when the frame motion mode is previous image selection mode.

41. The method for generating motion descriptors by using accumulated motion histogram according to claim 38, wherein said step of motion vector values comparing/converting comprises comparing the absolute values of the computed differences between the present first area and the present second area and between the present first area and the present third area, and respectively converting the input X, Y motion vector values into 0 if all the absolute values are smaller than a predetermined threshold value, when the frame motion mode is present image selection mode.

42. The method for generating motion descriptors by using accumulated motion histogram according to claim 41, wherein said step of motion vector values comparing/converting comprises providing Y without any conversion and 0 after converting the input X motion vector values into 0, if the absolute value of the computed difference between the present first area and the present second area and the absolute value of the computed difference between the present first area and the present third area are not commonly smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the present third area is smaller than the predetermined threshold value.

43. The method for generating motion descriptors by using accumulated motion histogram according to claim 41, wherein said step of motion vector values comparing/converting comprises providing X motion vector value without any conversion and 0 after converting the input Y motion vector value into 0, if the absolute value of the computed difference between the present first area and the present second area and the absolute value of the computed difference between the present first area and the present third area are not commonly smaller than the predetermined threshold value, if the absolute value of the difference between the present first area and the present third area is not smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the present second area is smaller than the predetermined threshold value.

44. The method for generating motion descriptors by using accumulated motion histogram according to claim 41, wherein said step of motion vector values comparing/converting comprises providing X, Y motion vector values without any conversion, if the absolute value of the computed difference between the present first area and the present second area and the absolute value of the computed difference between the present first area and the present third area are not smaller than the predetermined threshold value, if the absolute value of the computed difference between the present first area and the present third area is not smaller than the predetermined threshold value, or if the absolute value of the computed difference between the present first area and the present second area is smaller than the predetermined threshold value.

45. The method for generating motion descriptors by using accumulated motion histogram according to claim 38, wherein said step of motion vector values comparing/converting comprises comparing the absolute value of the computed difference between the present first area and the previous first area, which is computed in said step of absolute value computing, and converting the respective input X, Y motion vector values into 0 if the absolute value is smaller than a predetermined threshold value.

46. The method for generating motion descriptors by using accumulated motion histogram according to claim 45, wherein said step of motion vector values comparing/converting provides X, Y motion vector values after respectively converting them into 0, if the absolute value of the computed difference between the present first area and the previous first area is not smaller than the predetermined threshold value, or if all the absolute values of the computed differences between the present first area and the previous second area and between the present first area and the previous third area are commonly smaller than the predetermined threshold value.

47. The method for generating motion descriptors by using accumulated motion histogram according to claim 45, wherein said step of motion vector values comparing/converting provides Y motion vector value without any conversion and 0 after converting the input X motion vector value into 0 if the absolute value of the computed difference between the present first area and the previous first area, if all the absolute values of the computed differences between the present first area and the previous second area and between the present first area and the previous third area are not smaller than the predetermined threshold value, or if the absolute value of the computed difference between the present first area and the previous third area is smaller than the predetermined threshold value.

48. The method for generating motion descriptors by using accumulated motion histogram according to claim 45, wherein said step of motion vector values comparing/converting provides X motion vector value without any conversion and 0 after converting the input Y motion vector value into 0, if the absolute value of the computed difference between the present first area and previous first area is not smaller than the predetermined threshold value, if all the absolute values of the computed differences between the present first area and the present second area and between the present first area and the present third area are not smaller than the predetermined threshold value, if the absolute value of the computed difference between the present first area and the present third area is not smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the previous second area is smaller than the predetermined threshold value.

49. The method for generating motion descriptors by using accumulated motion histogram according to claim 45, wherein said step of motion vector values comparing/converting provides X, Y motion vector values without any conversion, if the absolute value of the computed difference between the present first area and previous first area is not smaller than the predetermined threshold value, if all the absolute values of the computed differences between the present first area and the previous second area and between the present first area and the previous third area are not smaller than the predetermined threshold value, if the absolute value of the computed difference between the present first area and the previous third area is not smaller than the predetermined threshold value, or if the absolute value of the difference between the present first area and the present second area is smaller than the predetermined threshold value.

50. The method for generating motion descriptors by using accumulated motion histogram according to claim 36 or claim 37, wherein said step of motion filtering comprises the steps of:

motion intensity computing for computing motion intensity by using the X,Y motion vector values which are processed by said step of motion estimation value diverging;

motion vector value comparing/converting for comparing motion intensity which is computed by said motion intensity computation part with a predetermined threshold value, and performing conversion of the motion vectors according to the comparison result;

motion direction computing for computing motion direction by using the motion vector values which are converted by said step of motion vector value comparing/converting part; and

motion direction quantizing/inverse-quantizing for quantizing/inverse-quantizing motion direction which is computed by said step of motion direction computation.

51. The method for generating motion descriptors by using accumulated motion histogram according to claim 50, wherein said step of motion vector values comparing/converting comprises converting X,Y motion vector values, which are outputted from said step of motion estimation value diverging, into 0, when the motion intensity which is computed in said step of motion intensity computing is smaller than a predetermined value, and providing X,Y motion vector values, which are outputted from said step of motion estimation value diverging, without any conversion, when the motion intensity which is computed in said step of motion intensity computing is larger than the predetermined value.

52. The method for generating motion descriptors by using accumulated motion histogram according to claim 50, wherein in said step of motion direction quantizing/inverse-quantizing, quantization of the motion direction, which is computed by said motion direction computation part, is performed by using either a linear quantization method or a non-linear quantization method.

69. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, wherein the video ID descriptor is to describe a credit for identifying video name, video producer, source and offerer, production time of the video and version of the video.

70. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, wherein the time descriptor is divided into a sequential time descriptor and a random time descriptor according to whether the motion descriptor represents all motion data in the description scope or not.

71. The method for generating motion descriptors by using accumulated motion histogram according to claim 69, wherein the sequential time descriptor is to describe starting time, ending time, and duration of a motion data.

72. The method for generating motion descriptors by using accumulated motion histogram according to claim 69, wherein the random time descriptor is to describe partial motion data and characteristics in a description time.

73. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, wherein the motion direction descriptor is to describe a motion direction average (θ

1), temporal distribution of motion direction (θ

p), deviation average of motion direction (σ

θ

1), spatial distribution of motion direction (σ

θ

p), dominant motion direction, accumulated histogram of motion direction, and motion intensity average of motion direction.

74. The method for generating motion descriptors by using accumulated motion histogram according to claim 73, wherein the motion direction average (θ

1) represents an average of motion direction averages (ρ

) of T images in a scope which is intended by the time descriptor in the motion descriptor, which is obtained by an equation as;

${\mathrm{\theta <br><br>}}_1=\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{\rho <br><br>}}_i/T$

wherein T may not equal to numbers of all images intended by the time descriptor.

75. The method for generating motion descriptors by using accumulated motion histogram according to claim 73, wherein the temporal distribution of the motion direction represents characteristics of temporal distribution and distortion degree with relation to the direction averages (ρ

) of respective images in the direction average (θ

1), which is obtained by an equation as;

${\mathrm{\theta <br><br>}}_p=\sqrt{\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\left({\mathrm{\rho <br><br>}}_i-\mathrm{\theta <br><br>}\right)}^p/T},\text{\hspace{1em}<br><br>}<br><br>P\ge <br><br>2.$

76. The method for generating motion descriptors by using accumulated Lotion histogram according to claim 73, wherein the deviation average of the motion direction represents an average of the spatial distribution of directions with relation to total T Images in the time, which is obtained by an equation as:

${\mathrm{\sigma <br><br>}}_{\mathrm{\theta <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>1}=\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{\sigma <br><br>}}_{\mathrm{i1}}/T.$

77. The method for generating motion descriptors by using accumulated motion histogram according to claim 73, wherein the spatial distribution of motion direction represents characteristics of spatial distribution and distortion with relation to standard deviations (σ

τ

i) of respective images in standard deviation averages (σ

θ

1), which is obtained by an equation as;

${\mathrm{\sigma <br><br>}}_{\mathrm{\theta <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>p}=\sqrt{\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>\left({\mathrm{\sigma <br><br>}}_{\mathrm{i1}}-{\mathrm{\sigma <br><br>}}_{\mathrm{\theta <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>1}\right)/T},\text{\hspace{1em}<br><br>}<br><br>P\ge <br><br>2.$

78. The method for generating motion descriptors by using accumulated motion histogram according to claim 73, wherein the dominant direction of motion represents a direction of which histogram value is largest among motion histograms which are generated with relation to direction by using below formula (5), wherein a plurality of dominant directions may be determined in sequence of the histogram value:

79. The method for generating motion descriptors by using accumulated motion histogram according to claim 9, wherein the intensity descriptor Is to describe average of motion intensity (ω

1), temporal distribution of motion intensity (ω

q), deviation average of motion intensity (σ

ω

q), spatial distribution of motion intensity (σ

ω

q), accumulated histogram of motion intensity (Hω

), and average of motion intensity with relation to motion direction.

80. The method for generating motion descriptors by using accumulated motion histogram according to claim 79, wherein the motion intensity average (ω

1) represents an average of motion intensity averages (λ

) of T times in a scope which is intended by the time descriptor in the motion descriptor, which is obtained by an equation as;

${\mathrm{\omega <br><br>}}_1=\left(\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{\lambda <br><br>}}_i\right)/T$

wherein T may not equal to numbers of all images in the description time.

81. The method for generating motion descriptors by using accumulated motion histogram according to claim 79, wherein the temporal distribution of motion intensity (ω

q) is obtained by an equation as;

${\mathrm{\omega <br><br>}}_q=\sqrt{\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\left({\mathrm{\lambda <br><br>}}_t-{\mathrm{\omega <br><br>}}_1\right)}^q/T},\text{\hspace{1em}<br><br>}<br><br>q\ge <br><br>2.$

82. The method for generating motion descriptors by using accumulated motion histogram according to claim 79, wherein the deviation average of the motion direction (σ

ω

1) represents an average of the spatial distribution of intensity with relation to total T images in the time, which is obtained by an equation as;

${\mathrm{\sigma <br><br>}}_{\mathrm{\omega <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>1}=\left(\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{\sigma <br><br>}}_{\mathrm{\lambda <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>i}\right)/T.$

83. The method for generating motion descriptors by using accumulated motion histogram according to claim 79, wherein the spatial distribution of motion intensity (σ

ω

q) is obtained by an equation as;

${\mathrm{\sigma <br><br>}}_{\mathrm{\omega <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>q}=\sqrt{\underset{i=0}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\left({\mathrm{\sigma <br><br>}}_{\mathrm{\lambda <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>i}-{\mathrm{\sigma <br><br>}}_{\mathrm{\omega <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>1}\right)}^q/T},\text{\hspace{1em}<br><br>}<br><br>q\ge <br><br>2.$