On The Content Based Animated Video Retrieval Technology Research

In order to make the majority of animation enthusiasts to be able to quickly extractinteresting animated video to watch, but also to many animators in the animationvideo library can find what they want faster and better material, and news videos,sports video, advertising video and other digital video, animated video is also facedwith the need for content-based video retrieval.After analyzing the animated video of the hierarchy on the basis of the highlightsthat are involved in the retrieval process of the animated video shot boundarydetection, keyframe extraction, as well as animation division of the story. Firstly, wedetect the animated video shot boundary with the lens as the basic structural unit.Secondly, we extract the key frame from each lens segment and use it as the shotcontent, namely, we use a more comprehensive, accurate, concise summary of thekey frames as the lens main content, then through the lens clustering method toachieve animation plot division in order to abbreviate the animated content.In this paper, the characteristics of the animated video shot boundary detectionusing adaptive threshold boundary detection algorithm based on interframedifference between non-adjacent interframe difference and made someimprovements to the algorithm.It also presents a content based adaptive thresholdanalysis method in the key frame extraction. Story detection using an improvedK-means algorithm based on the shot key frames to the division of the story of theanimated video. The algorithm used in this article has been tested through theexperiments. The results show that the adopted algorithm can achieve the betteranimation story division and express accurate, abbreviated animation content.