Video Summary Of Research And Systematic Implementation

As the fast development of network and the wide application of multimedia technology, a mass of digital videos have arisen. How to browse the large volume of video data and how to attain and express the contents of videos has become a problem urgent to solve. So the video abstraction technology becomes the people's focus. The video scene cutting and the key frame abstraction are the two key technologies in the video abstraction. This paper first introduces the research background, development progress and current researches, and then summaries the traditional video abstraction technologies, and studies the advantages and disadvantages of the current video cutting and key frame abstraction methods.According to the summary of traditional video scene cutting algorithms, the paper proposes a new scene cutting method based the HSV color model space. This method transforms the RGB values of each image to the HSV values first, and quantizes to get the one-dimensional eigenvector. After calculating the histogram, we can get the similarity of two pictures. And then comparing the similarity and the threshold can we judge if the scene has switched. The experiment has proved that this method can solve the problem that two different pictures may not be distinguished because the same RGB histograms. It can help to cut the video scenes correctly.This paper also puts forward a key frame abstraction method based twice-clusters after analyzing the advantages and disadvantages of traditional key frame abstraction algorithms. It imports the segmentation technology when calculating the similarities. It can show the key contents of video pictures by given different coefficients to different segments. Calculating the adaptive threshold value can help us do the first-cluster to the videos. And then count the distances between two classes, and get the final classes after the second-cluster. The pictures nearest the class center in each class are the key frames. The method can overcome the disadvantages that the key frames will be redundant if only once-cluster. The experiment has showed that the key frames can embody the video contents compressively and correctly.Based on the video scene cutting and key frame abstraction algorithms, we develop a simple video abstraction system aiming at AVI video files after studying the physicalstructure. It contains several function modules such as video playing, scene cutting,key frame abstracting and abstraction displaying. The system can satisfy the users'basic requirements.In the end, the paper sums up the video abstraction technology and the experimentsystem, and presents the following works.