Development of content based video retrieval CBVR system for searching similar videos in real time applications

Abstract

With the emergence of Multimedia and other high-end applications newlineon the web, smartphones and other digital technologies, there is tremendous newlineusage of images and videos. Videos and images influence all walks of life. newlineContent-based Image and Video Search has become an interesting task in the newlinefield of Computer Vision, to search for the desired video from a large collection newlineof videos. Due to the fast growth in the sales of smartphones, CCTV and newlinecameras, the size of video content on the web has increased by leaps and newlinebounds. This makes the manual annotation process a time-consuming and newlinetedious process. Instead of browsing hours for videos for a particular scene newlinesequence, it can be searched quickly by specifying any video frame present in newlinethe continuous video. Detecting a specific video from numerous videos is a newlinesignificant issue in numerous applications like video search, video surveillance, newlineadvertisement discovery, clinical diagnosis, crime identification, etc. newlineTherefore, developing Realtime Content-Based Video Retrieval newline(CBVR) that searches for similar videos becomes inevitable. This effort can newlinereduce the browsing time of video users and improve the accuracy of the video newlinesearch. Hence, this work focuses on improving the accuracy and search newlineresponse time of the video search process. newlineContent-Based Video Retrieval for searching similar videos in realtime newlineapplications is a distributed system that can search similar videos based on newlinethe user input with high accuracy and low search response time. The presented newlinesystem forms a video processing pipeline, consisting of three major newlinecontributions. newline