CN109361923A - A kind of time slip-window scene change detection method and system based on motion analysis - Google Patents

Abstract

The present invention proposes a kind of time slip-window scene change detection method and system based on motion analysis.The method of the present invention passes through the analysis of the exercise intensity of object in image first, the adaptive length for determining scene analysis time window；Then as unit of the time window, cycle detection is done to scene switching.In this way under the accuracy for guaranteeing scene detection, while unnecessary waste of the detection algorithm in calculation amount frame by frame is avoided, to achieve the purpose that boosting algorithm efficiency.

Description

Sliding time window scene switching detection method and system based on motion analysis

Technical Field

The invention relates to the field of video processing, in particular to a method and a system for detecting scene switching of a sliding time window based on motion analysis.

Background

The actual video source has diversity, and the compression performance of the encoder can be improved by accurate scene switching. However, although the conventional frame-by-frame analysis method is accurate in judgment, the probability of scene switching is actually extremely low frame-by-frame, which causes the inefficiency of the algorithm.

Disclosure of Invention

The embodiment of the invention aims to provide a sliding time window scene switching detection method based on motion analysis, and aims to solve the problem that the scene switching frame-by-frame analysis method in the prior art is accurate in judgment but low in efficiency.

The embodiment of the invention is realized in such a way that a method for detecting scene switching of a sliding time window based on motion analysis comprises the following steps:

step 1: if the current frame_curIf the frame is an intra-frame prediction frame or a scene switching frame, the cyclic variable n is given an initial value of 0, and the frame is determinedThe next frame in coding order of the current frame is marked as frame_nextThen go to Step 2; otherwise, go to Step 6;

step 2: if frame_cur、frame_nextIf there is an uncoded frame, it is coded first, and then Step3 is entered; otherwise, go directly to Step 3;

step 3: if poc_next-poc_cur≤Thres₁Then go to Step 5; otherwise, go to Step 4;

step 4: if it is notThen the frame is decided_nextFrame switching for scenes_cur＝frame_nextAnd update the frame_nextThen go back to Step 2; otherwise, go to Step 5;

step 5: acquiring a time span variable tk of a current scene;

step 6: carrying out scene detection on the video residual frame;

wherein, poc_cur、poc_nextRespectively representing frames_cur、frame_nextThe play sequence number of (1); thres₁Represents a first threshold value;representing a frame_nextThe number of intra prediction blocks involved; thres₂Representing a second threshold value.

Another objective of an embodiment of the present invention is to provide a system for detecting scene switching in a sliding time window based on motion analysis, where the system includes:

a frame type judgment processing module for judging if the frame of the current frame is_curIf the frame is an intra-frame prediction frame or a scene switching frame, the cyclic variable n is given an initial value of 0, and the next frame of the current frame in the coding sequence is positioned and recorded as a frame_nextThen enters the detection site of the uncoded frameA management module; otherwise, entering a residual frame scene detection device;

an uncoded frame detection processing module for determining if frame_cur、frame_nextIf there is an uncoded frame, it is coded first, and then enters a frame interval threshold judgment processing module; otherwise, directly entering a frame interval threshold value judging and processing module;

a frame interval threshold judgment processing module for judging if poc_next-poc_cur≤Thres₁Entering a scene time span variable acquisition module; otherwise, entering an intra-frame prediction block quantity threshold value judgment processing module;

wherein, poc_cur、poc_nextRespectively representing frames_cur、frame_nextThe play sequence number of (1); thres₁Represents a first threshold value;

a threshold judgment processing module for judging ifThen the frame is decided_nextFrame switching for scenes_cur＝frame_nextAnd update the frame_nextThen reentering the uncoded frame detection processing module; otherwise, entering a scene time span variable acquisition module;

wherein,representing a frame_nextIncluding the number of intra prediction blocks; thres₂Represents a second threshold value;

the scene time span variable acquisition module is used for acquiring a current scene time span variable;

and the residual frame scene detection device is used for carrying out scene detection on the video residual frame.

The invention has the advantages of

The invention provides a sliding time window scene switching detection method and system based on motion analysis. The method comprises the steps of firstly, adaptively determining the length of a scene analysis time window through the analysis of the motion intensity of an object in an image; and then, taking the time window as a unit, and carrying out periodic detection on scene switching. Therefore, under the condition of ensuring the accuracy of scene detection, the meaningless waste of the frame-by-frame detection algorithm on the calculated amount is avoided, and the purpose of improving the algorithm efficiency is achieved.

Drawings

FIG. 1 is a flow chart of a method for detecting scene cuts in a sliding time window based on motion analysis according to a preferred embodiment of the present invention;

FIG. 2 is a flowchart of the detailed method of Step6 in FIG. 1;

FIG. 3 is a diagram of a system for detecting scene cuts in a sliding time window based on motion analysis according to a preferred embodiment of the present invention;

fig. 4 is a block diagram of the residual frame scene detection apparatus in fig. 3.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and examples, and for convenience of description, only parts related to the examples of the present invention are shown. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The invention provides a sliding time window scene switching detection method and system based on motion analysis. The method comprises the steps of firstly, adaptively determining the length of a scene analysis time window through the analysis of the motion intensity of an object in an image; and then, taking the time window as a unit, and carrying out periodic detection on scene switching. Therefore, under the condition of ensuring the accuracy of scene detection, the meaningless waste of the frame-by-frame detection algorithm on the calculated amount is avoided, and the purpose of improving the algorithm efficiency is achieved.

Example one

FIG. 1 is a flow chart of a method for detecting scene cuts in a sliding time window based on motion analysis according to a preferred embodiment of the present invention; the method comprises the following steps:

step 1: if the current frame_curIf the frame is an intra-frame prediction frame or a scene switching frame, the cyclic variable n is given an initial value of 0, and the next frame of the current frame in the coding sequence is positioned and recorded as a frame_nextThen go to Step 2; otherwise, Step6 is entered.

Step 2: if frame_cur、frame_nextIf there is an uncoded frame, it is coded first, and then Step3 is entered; otherwise, go directly to Step 3.

Step 3: if poc_next-poc_cur≤Thres₁Then go to Step 5; otherwise, go to Step 4.

Wherein, poc_cur、poc_nextRespectively representing frames_cur、frame_nextThe play sequence number of (1); thres₁Representing a first threshold value, typically Thres₁＞2。

Step 4: if it is notThen the frame is decided_nextFrame switching for scenes_cur＝frame_nextAnd update the frame_nextThen go back to Step 2; otherwise, go to Step 5.

Wherein,representing a frame_nextIncluding the number of intra prediction blocks; thres₂Indicating a second thresholdValue, typically Thres₂≥0.9。

Step 5: acquiring a time span variable of a current scene;

wherein, cons_fast、cons_slowAnd cons respectively represent a fast scene time span constant, a slow scene time span constant and a conventional scene time span constant, and the common value is 1 < cons_fast＜cons＜cons_slow＜fps/2；Thres₃、Thres₄Representing a third and a fourth threshold value; fps represents a frame rate; bit_next、bit_curRespectively representing frames_cur、frame_nextThe number of coded bits; the value division condition of the time span variable can also be frame_nextThe occupation ratio of the non-zero blocks of the medium motion vector is used for operation; the larger the time span constant is set, the faster the processing speed is, but a certain omission factor is also brought, and the setting can be carried out according to specific requirements.

Step 6: and carrying out scene detection on the video residual frame.

FIG. 2 is a flowchart of the detailed method of Step6 in FIG. 1; the method comprises the following steps:

step 61: let the loop variable n be n + 1.

Step 62: positioning the frame to be detected of the current scene according to the time span variable_t，

t＝poc_next+tk*n。

Step 63: if the frame to be detected does not exist in the current scene, ending the process; otherwise, if the frame to be detected in the current scene is the I frame and the interval between the I frames is a fixed value, the frame is made_cur＝frame_tThen go back to Step 1; otherwise, Step64 is entered.

Wherein the frame is_tIndicating the current scene to be detected frame, t indicating its playing sequenceNumber; the I frame interval is preset in the encoding parameters and is a fixed value.

Step 64: for frames using methods disclosed in the art_tAnd carrying out scene detection.

Step 65: order frame_cur＝frame_tAnd then go back to Step 1.

Example two

FIG. 3 is a diagram of a system for detecting scene cuts in a sliding time window based on motion analysis according to a preferred embodiment of the present invention; the system comprises:

a frame type judgment processing module for judging if the frame of the current frame is_curIf the frame is an intra-frame prediction frame or a scene switching frame, the cyclic variable n is given an initial value of 0, and the next frame of the current frame in the coding sequence is positioned and recorded as a frame_nextThen, the coded frame enters an uncoded frame detection processing module; otherwise, the method enters a residual frame scene detection device.

An uncoded frame detection processing module for determining if frame_cur、frame_nextIf there is an uncoded frame, it is coded first, and then enters a frame interval threshold judgment processing module; otherwise, directly entering a frame interval threshold value judging and processing module.

A frame interval threshold judgment processing module for judging if poc_next-poc_cur≤Thres₁Entering a scene time span variable acquisition module; otherwise, entering an intra-frame prediction block quantity threshold value judgment processing module.

Wherein, poc_cur、poc_nextRespectively representing frames_cur、frame_nextThe play sequence number of (1); thres₁Representing a first threshold value, typically Thres₁＞2。

A threshold judgment processing module for judging ifThen the frame is decided_nextFrame switching for scenes_cur＝frame_nextAnd update the frame_nextThen reentering the uncoded frame detection processing module; otherwise, entering a scene time span variable acquisition module.

Wherein,representing a frame_nextIncluding the number of intra prediction blocks; thres₂Representing a second threshold, typically Thres₂0.9 of.

The scene time span variable acquisition module is used for acquiring a current scene time span variable;

wherein, cons_fast、cons_slowAnd cons respectively represent a fast scene time span constant, a slow scene time span constant and a conventional scene time span constant, and the common value is 1 < cons_fast＜cons＜cons_slow＜fps/2；Thres₃、Thres₄Representing a third and a fourth threshold value; fps represents a frame rate; bit_next、bit_curRespectively representing frames_cur、frame_nextThe number of coded bits; the value division condition of the time span variable can also be frame_nextThe occupation ratio of the non-zero blocks of the medium motion vector is used for operation; the larger the time span constant is set, the faster the processing speed is, but a certain omission factor is also brought, and the setting can be carried out according to specific requirements.

And the residual frame scene detection device is used for carrying out scene detection on the video residual frame.

Fig. 4 is a block diagram of the residual frame scene detection apparatus in fig. 3. The residual frame scene detection device includes:

and the loop variable setting module is used for enabling the loop variable n to be n + 1.

A scene frame positioning module for positioning the frame to be detected of the current scene according to the time span variable_t，t＝poc_next+tk*n。

The first judgment processing module is used for judging whether the frame to be detected of the current scene does not exist or not, and ending the judgment; otherwise, entering a second judgment processing module;

a second judgment processing module for judging if the frame to be detected in the current scene is an I frame and the inter-frame interval is a fixed value, then making the frame_cur＝frame_tThen returning to the frame type judgment processing module; otherwise, entering a scene detection module.

Wherein the frame is_tRepresenting a frame to be detected of a current scene, and t representing a playing sequence number of the frame; the I frame interval is preset in the encoding parameters and is a fixed value.

A scene detection module for frame using methods disclosed in the industry_tAnd carrying out scene detection.

A current frame setting module for ordering the frame_cur＝frame_tThen, the frame type judgment processing module is returned.

It will be understood by those skilled in the art that all or part of the steps in the method according to the above embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer readable storage medium, such as ROM, RAM, magnetic disk, optical disk, etc.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (8)

1. A method for detecting scene switching of a sliding time window based on motion analysis is characterized by comprising the following steps:

step 1: if the current frame_curIf the frame is an intra-frame prediction frame or a scene switching frame, the cyclic variable n is given an initial value of 0, and the next frame of the current frame in the coding sequence is positioned and recorded as a frame_nextThen go to Step 2; otherwise, go to Step 6;

step 2: if frame_cur、frame_nextIf there are any uncoded frames, then it is coded first and then St is enteredep 3; otherwise, go directly to Step 3;

step 3: if poc_next-poc_cur≤Thres₁Then go to Step 5; otherwise, go to Step 4;

step 4: if it is notThen the frame is decided_nextFrame switching for scenes_cur＝frame_nextAnd update the frame_nextThen go back to Step 2; otherwise, go to Step 5;

step 5: acquiring a time span variable tk of a current scene;

step 6: carrying out scene detection on the video residual frame;

wherein, poc_cur、poc_nextRespectively representing frames_cur、frame_nextThe play sequence number of (1); thres₁Represents a first threshold value;representing a frame_nextThe number of intra prediction blocks involved; thres₂Representing a second threshold value.

2. The method of motion analysis based sliding time window scene cut detection according to claim 1,

first threshold value Thres₁Is more than 2; second threshold value Thres₂≥0.9。

3. The method for detecting scene cut with sliding time window based on motion analysis according to claim 1, wherein the obtaining the current scene time span variable tk is specifically:

wherein, cons_fast、cons_slowCons respectively represent fast scene time span constantsSlow scene time span constant, conventional scene time span constant, 1 < cons_fast＜cons＜cons_slow＜fps/2；Thres₃、Thres₄Representing a third and a fourth threshold value; fps represents a frame rate; bit_next、bit_curRespectively representing frames_cur、frame_nextThe number of coded bits.

4. The motion-analysis-based sliding-time-window scene-cut detection method of claim 1, wherein the scene detection of the video residual frame comprises the steps of:

step 61: let the cyclic variable n be n + 1;

step 62: positioning the frame to be detected of the current scene according to the time span variable_t，

t＝poc_next+tk*n；

Step 63: if the frame to be detected does not exist in the current scene, ending the process; otherwise, if the frame to be detected in the current scene is the I frame and the interval between the I frames is a fixed value, the frame is made_cur＝frame_tThen go back to Step 1; otherwise, go to Step 64;

wherein the frame is_tRepresenting a frame to be detected of a current scene, and t representing a playing sequence number of the frame; the I frame interval is preset in the coding parameters and is a fixed value;

step 64: for frames using methods disclosed in the art_tCarrying out scene detection;

step 65: order frame_cur＝frame_tAnd then go back to Step 1.

5. A sliding time window scene cut detection system based on motion analysis, the system comprising:

a frame type judgment processing module for judging if the frame of the current frame is_curIf the frame is an intra-frame prediction frame or a scene switching frame, the cyclic variable n is given an initial value of 0, and the next frame of the current frame in the coding sequence is positioned and recorded as a frame_nextThen, the coded frame enters an uncoded frame detection processing module; otherwise, go to the restA residual frame scene detection device;

an uncoded frame detection processing module for determining if frame_cur、frame_nextIf there is an uncoded frame, it is coded first, and then enters a frame interval threshold judgment processing module; otherwise, directly entering a frame interval threshold value judging and processing module;

a frame interval threshold judgment processing module for judging if poc_next-poc_cur≤Thres₁Entering a scene time span variable acquisition module; otherwise, entering an intra-frame prediction block quantity threshold value judgment processing module;

wherein, poc_cur、poc_nextRespectively representing frames_cur、frame_nextThe play sequence number of (1); thres₁Represents a first threshold value;

a threshold judgment processing module for judging ifThen the frame is decided_nextFrame switching for scenes_cur＝frame_nextAnd update the frame_nextThen reentering the uncoded frame detection processing module; otherwise, entering a scene time span variable acquisition module;

wherein,representing a frame_nextIncluding the number of intra prediction blocks; thres₂Represents a second threshold value;

the scene time span variable acquisition module is used for acquiring a current scene time span variable;

and the residual frame scene detection device is used for carrying out scene detection on the video residual frame.

6. The motion analysis based sliding time window scene cut detection system of claim 5,

first threshold value Thres₁Is more than 2; second threshold value Thres₂≥0.9。

7. The system according to claim 6, wherein the scene time span variable acquiring module is configured to acquire the current scene time span variable as:

wherein, cons_fast、cons_slowCons respectively represents a fast scene time span constant, a slow scene time span constant and a conventional scene time span constant, 1 < cons_fast＜cons＜cons_slow＜fps/2；Thres₃、Thres₄Representing a third and a fourth threshold value; fps represents a frame rate; bit_next、bit_curRespectively representing frames_cur、frame_nextThe number of coded bits.

8. The motion analysis based sliding time window scene cut detection system of claim 7, wherein the residual frame scene detection means comprises:

a loop variable setting module, configured to set a loop variable n to n + 1;

a scene frame positioning module for positioning the frame to be detected of the current scene according to the time span variable_t，t＝poc_next+tk*n；

The first judgment processing module is used for judging whether the frame to be detected of the current scene does not exist or not, and ending the judgment; otherwise, entering a second judgment processing module;

a second judgment processing module for judging if the frame to be detected in the current scene is an I frame and the inter-frame interval is a fixed value, then making the frame_cur＝frame_tThen returning to the frame type judgment processing module; otherwise, entering a scene detection module;

wherein the frame is_tRepresenting a frame to be detected of a current scene, and t representing a playing sequence number of the frame; i frame spacingThe coding parameters are preset, and the interval between I frames is a fixed value;

a scene detection module for frame using methods disclosed in the industry_tCarrying out scene detection;

a current frame setting module for ordering the frame_cur＝frame_tThen, the frame type judgment processing module is returned.