Accurate detection of a drastic change between successive pictures

• US 5,032,905 A
• Filed: 05/11/1990
• Issued: 07/16/1991
• Est. Priority Date: 05/11/1989
• Status: Expired due to Fees

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Koga, Toshio
• Primary Examiner(s): Kostak, Victor R.
• Application Number:
  US07/522,026
• Time in Patent Office: 431 Days
• Field of Search: 358/105, 358/136, 358/135, 358/140
• US Class Current: 348/700
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: H04N 19/142 Detection of scene cut or scene change
• CPC Class Codes: H04N 19/179 the unit being a scene or a shot
• CPC Class Codes: H04N 19/87 involving scene cut or scene change detection in combination with video compression
• CPC Class Codes: H04N 5/144 Movement detection for video coding H04N19/503; analysis of motion in general G06T7/20
• CPC Class Codes: H04N 5/147 Scene change detection

Abstract

On detecting a drastic change, such as a scene change, between current and previous ones of successive pictures which are represented by a digital video signal and each of which comprises a predetermined number of picture elements, the digital video signal is processed by using correlations between the picture elements within each of the successive pictures to produce a processed signal which represents processed results for each of the successive pictures. A statistical distribution of the processed results is calculated from the processed signal. The statistical distribution of the processed results in the current picture is collated with the statistical distribution of the processed results in the previous picture to be produced as a change detection signal representative of the drastic change with reference to a relationship between the statistical distribution of the processed results in the current picture and the statistical distribution of the processed results in the previous picture. The processed signal may be obtained by producing a difference signal between the digital video signal and a prediction signal predictive of the digital video signal by using the correlations. Alternatively, an orthogonal transform is carried out for the digital video signal to produce the processed signal.

Claims

1. A method of detecting a scene change between current and previous pictures which are two adjacent ones of successive pictures represented by an input digital video signal, each of said successive pictures comprising a predetermined number of picture elements, said method comprising the steps of:

processing said input digital video signal by using correlations between the picture elements within each of said successive pictures to produce a processed signal which represents processed results for each of said successive pictures;

calculating a statistical distribution of said processed results in response to said processed signal; and

collating the statistical distribution of said processed results in said current picture with the statistical distribution of said processed results in said previous picture to produce a change detection signal representative of said scene change with reference to a relationship between the statistical distribution of said processed results in said current picture and the statistical distribution of said processed results in said previous picture.

2. A method as claimed in claim 1, wherein said processing step comprises the steps of:

predictively producing, in response to said input digital video signal, a prediction signal predictive of said input digital video signal by using correlations between the picture elements within each of said successive pictures; and

producing, as said processed signal, a difference signal which represents, as said processed results, differences between said input digital video signal and said prediction signal for each of said successive pictures.

3. A method as claimed in claim 1, wherein said processing step is for carrying out an orthogonal transform of said input digital video signal to produce, as said processed signal, a result signal which represents, as said processed results, transform results of said orthogonal transform for each of said successive pictures.

4. A method as claimed in claim 1, wherein said collating step comprises the steps of:

calculating distribution differences between the statistical distribution of said processed results in said current picture and the statistical distribution of said processed results in said previous picture; and

comparing each of said distribution differences with a predetermined threshold value to produce said change detection signal with reference to a relationship between each of said distribution differences and said predetermined threshold value.

5. A method as claimed in claim 4, wherein said distribution difference calculating step comprises the steps of:

temporarily memorizing the statistical distribution of said processed results in said current picture during one frame of said input digital video signal to produce a delayed distribution having a delay which is equal to said one frame; and

producing differences between the statistical distribution of said processed results in said current picture and said delayed distribution as said distribution differences.

6. A method as claimed in claim 4, wherein:

said distribution difference calculating step is for calculating a sum of absolute values of said distribution differences;

said comparing step being for comparing said sum with said predetermined threshold value to produce said change detection signal when said sum is greater than said predetermined threshold value.

7. A detecting device for detecting a scene change between current and previous pictures which are two adjacent ones of successive pictures represented by an input digital video signal, each of said successive pictures comprising a predetermined number of picture elements, said device comprising:

processing means for processing said input digital video signal by using correlations between the picture elements within each of said successive pictures to produce a processed signal which represents processed results for each of said successive pictures;

statistical distribution calculating means connected to said processing means and supplied with said processed signal for calculating a statistical distribution of said processed results; and

collating means connected to said statistical distribution calculating means for collating the statistical distribution of said processed results in said current picture with the statistical distribution of said processed results in said previous picture to produce a change detection signal representative of said scene change with reference to a relationship between the statistical distribution of said processed results in said current picture and the statistical distribution of said processed results in said previous picture.

8. A detecting device as claimed in claim 7, wherein said processing means comprises:

prediction signal producing means responsive to said input digital video signal for predictively producing a prediction signal predictive of said input digital video signal by using correlations between the picture elements within each of said successive pictures; and

difference signal producing means connected to said prediction signal producing means for producing, as said processed signal, a difference signal which represents, as said processed results, differences between said input digital video signal and said prediction signal for each of said successive pictures.

9. A detecting device as claimed in claim 7, wherein said processing means is for carrying out an orthogonal transform of said input digital video signal to produce, as said processed signal, a result signal which represents, as said processed results, transform results of said orthogonal transform for each of said successive pictures.

10. A detecting device as claimed in claim 7, wherein said collating means comprises:

distribution difference calculating means connected to said statistical distribution calculating means for calculating distribution differences between the statistical distribution of said processed results in said current picture and the statistical distribution of said processed results in said previous picture; and

comparing means connected to said distribution difference calculating means for comparing each of said distribution differences with a predetermined threshold value to produce said change detection signal with reference to a relationship between each of said distribution differences and said predetermined threshold value.

11. A detecting device as claimed in claim 10, wherein said distribution difference calculating means comprises:

temporary memorizing means connected to said statistical distribution calculating means for temporarily memorizing the statistical distribution of said processed results in said current picture during one frame of said input digital video signal to produce a delayed distribution having a delay which is equal to said one frame; and

means connected to said statistical distribution calculating means and to said temporary memorizing means for producing differences between the statistical distribution of said processed results in said current picture and said delayed distribution as said distribution differences.

12. A detecting device as claimed in claim 10, wherein:

said distribution difference calculating means is for calculating a sum of absolute values of said distribution differences;

said comparing means being for comparing said sum with said predetermined threshold value to produce said change detection signal when said sum is greater than said predetermined threshold value.