A machine learning method and system for intelligent filling of security monitoring video occlusion

Abstract

The present invention provides a machine-learning method and system for intelligent filling of security monitoring video occlusion, including: defining an N*M grid matrix template, and putting it into each frame of the security monitoring video that occludes a specific target In the video picture; calculate the feature quantity of the image area where each grid is located in the grid matrix template, combine the feature quantities of all the grids in the grid matrix template to obtain the grid matrix template An N*M grid matrix vector; perform clustering calculation on all grid matrix vectors of all video frames, and perform cluster division; determine the cluster where a specific target is located, and extract it according to the unoccluded part of the specific target in the cluster To fill the feature quantity of the occluded part of the specific target in each frame of video above. The invention fills up the specific target through the non-blocking part of the specific target in each video frame, and solves the problem of recognition failure when the specific target is continuously blocked.

Description

technical field [0001] The invention belongs to the technical field of video monitoring, and in particular relates to a machine-learning method and system for intelligent filling of security monitoring video occlusion. Background technique [0002] For security monitoring application scenarios, the recognition, extraction, and tracking of specific targets (such as specific people, specific vehicles) from video frames are a basic function. In the above recognition, extraction, and tracking process, first set a picture frame of an appropriate size, then intercept a part of the picture area with the picture frame from each frame of video frame, and extract feature quantities (such as contour features, color, etc.) from the picture area. feature, texture feature, etc.), and compare it with the feature threshold corresponding to the specific target. If the feature value is greater than the threshold, it indicates that the picture area where the frame is located is the position of...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the present invention provides a machine learning method and system for intelligent filling of security monitoring video occlusion. For a security monitoring video, when a specific target is continuously blocked by an occluder, the specific target in each video frame Non-occluded parts, fill in the specific target, and solve the problem of recognition failure when the specific target is continuously occluded

Images