Video content similarity discrimination method and system based on computer vision

Abstract

The embodiment of the invention discloses a video content similarity discrimination method and a system based on computer vision, and relates to the technical field of video image processing, and themethod comprises the following steps: respectively extracting key frames of a video A and a video B through an unsupervised learning clustering algorithm based on an HSV model; respectively calculating a hash value in each key frame of the video A and the video B through a perceptual hash algorithm; and based on the hash value in each key frame of the video A and the video B, judging the similarity of the video A and the video B through a Hamming distance algorithm. According to the method and the system, the problems that the representativeness of key frames is not enough, the accuracy can begreatly reduced under the condition that the content change amplitude between the frames is relatively small, and the effect of the existing picture local feature value processing technology is poorunder the condition of image brightness change are effectively solved; hash value comparison is carried out through the Hamming distance, similarity judgment is carried out on videos of different channels, and therefore the purpose of video copyright infringement judgment is achieved.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of video image processing, and in particular to a method and system for judging video content similarity based on computer vision. Background technique [0002] With the development of Web 2.0 and community-based networks, video multimedia has gradually become a hot spot for Internet applications. Users can shoot, edit, and upload video clips, which makes the number of videos rapidly expand. For example, a video sharing website has an average of tens of millions of new uploaded videos every day. Video infringements occur frequently, but it is difficult for video rights holders to find infringements. Therefore, how to identify infringing videos among these videos is an urgent problem to be solved in video applications. [0003] The video content similarity discrimination method is mainly based on video key frame extraction technology and image local feature value processing technology. ...

AI Technical Summary

Problems solved by technology

[0009] For this reason, the embodiment of the present invention provides a video content similarity discrimination method and system based on computer vision to solve the problem that the key frames extracted in the existing video key frame extraction technology are not representative enough, and the content variation between frames is small In the case of , the accuracy rate will be greatly reduced, the effect is not stable enough, and the existing image local eigenvalue processing technology ignores the image color information and the effect is not good when the image brightness changes.

Images