A method and system for face recognition video segment screening based on key frame backtracking

Abstract

The invention discloses a face recognition video clip screening method and system based on key frame backtracking. The present invention is mainly based on a composite detection method for multi-pose faces and a key frame backtracking face recognition method that can automatically construct a sample set, according to Based on the principle of modular design, a method of selecting specified character segments from videos is proposed, and a system of video analysis, screening, and segment selection is constructed. The invention can accurately select the video segment containing the designated person from the video stream to be tested, and satisfies the requirement of computer-aided manpower for video information screening under the background of information explosion era.

Description

technical field [0001] The invention belongs to the technical field of face recognition, and relates to a face recognition video clip screening method and system, in particular to a face recognition video clip screening method and system based on key frame backtracking. Background technique [0002] Face recognition refers to computer technology that uses analysis and comparison of facial visual information for identity identification. In general, a face recognition system includes image acquisition, image processing, face detection, and face recognition. [0003] Face detection is an important part of the face recognition process. The current face detection methods can be divided into knowledge-based methods, statistical model-based methods, and pattern-matching-based methods. [0004] At present, the commonly used methods in face recognition include geometric feature method, subspace analysis method, statistical feature method and so on. Despite great progress in the fi...

AI Technical Summary

Problems solved by technology

In these scenarios, although the accuracy of face recognition is very high, the user's degree of freedom is not high. Users can only use it to identify their own identity, and cannot choose other video streams or human objects to be detected.

For example, if a user wants to select a video clip containing a specified character from a work shot by himself, or a video clip containing a favorite actor from a film and television work, there is a lack of relevant application software.

Images