Research Of Shot Boundary Detection And Key Frame Extraction

With the rapid development of multimedia technology and computer network, It's easier for people to produce, store and transmit video information. How to organize and manage effectively these videos, and access the information quickly, has become a problem. Contented-Based Video Retrieval comes into being, and has become a hot spot of multimedia technology research. Shot boundary detection and key frame extraction are two key technologies.First, we present the background and significance of Contented-Based Video Retrieval technology. Then, we sum up the existing shot boundary detection methods, and analyze the advantages and disadvantages of each method. We find that the performance of cut detection is ideal, but gradual change detection is not satisfactory, the main reason is object motion in the shot affects the detection of gradual change border, leading to a number of false detection and reduce the performance of the algorithm.We present a novel method of shot boundary detection based on the knowledge of mutual information and canny edge detector. We extract video frame edge using canny edge detector, then distinguish object motion and shot transform effectively by analyzing frame edge differences, reducing error detections caused by object motion, improving recall and precision. Experiments prove this method is robust to object motion, performance is better.After the analysis of the traditional methods of key frame extraction, a new method of key frame extraction based on c1uster is presented. We first mark the shots of containing objects exercise, then use cluster technology to extract the key frame; On the others shots, we use the method based on shot boundary to extract key frame. This method overcomes the traditional algorithm disadvantage that key frame number of each shot is fixed. The experimental results show that the method of key frame extraction is an effective way to summarize video content, and improve the performance of the time.