SHOT SEGMENTATION FOR CONTENT BASED VIDEO RETRIEVAL

Content based video retrieval is a technique used to search and browse large collections of videos stored in a database. This technique has proven to be useful for numerous applications spreadacross different domains like, surveillance, security, biomedicine, and traffic regulation. Here, the analysis is carried out based on certain properties extracted from the video frames such as colour, edge, motion, and texture. Instead of storing the features of every single frame of the video, only the features of the representative frames,which describethe entire video, are stored. This results in better storage memory utilization. We propose a method which aims at efficiently segmenting the video into shots and selecting the key frames of each shot accordingly. In order to determine the shot boundaries, we have incorporated Colour Histogram and Background Subtraction methods in this paper.The analysis of the proposed technique is carried out for different videos