Multi Key Frame Abstraction from Videos Using Fuzzy Classifiers

With a vast number of stored videos in a video archive, it is not possible to produce by hand a trailer for every video stored. To organize and retrieve video information effectively has been a problem to be solved in the fields of database and information retrieval. The key frame extraction is a process which extracts the most representative image collections from the original video and is the basis of video analysis and retrieval. The objective of this paper is to generate multi-key frames by adding efficient Fuzzy C-Classifiers. A better and faster correlation maps can be generated to extract semantically meaningful information from the videos with overlapping views. Thus the goal is to retrieve the images meeting with specific visual feature descriptions from extensive video database, according to the features such as scenes, moving object in the video data, color, textures and shapes in the image data automatically without human involvement which facilitates quick browsing