Research And Implementation Of Information Entropy In Key Frame Extraction In Video Retrieval

Nowadays, multimedia has become the main way of information transmission and is widely used in all walks of life, such as advertising, education, medicine, business, entertainment and other. At the same time, along with the rapid development of computer and network technology, it becomes more and more convenient to transport information; in addition to radio, television and other traditional means of communication, mobile phones, tablet PCs and other devices have gradually become the mainstream of information dissemination. Currently, a lot of data will be produced everyday, and it has become more and more a urgent requirement to find effective data in a large number of information. Accurate video abstract can improve the efficiency of video retrieval. This thesis improves the accuracy of video summary by introducing the information entropy.In this thesis, the technology of video retrieval is studied, and the main work is as follows:1.This thesis introduces the research background and the present research status and the logical structure of the video. As different methods lead to different results and the same method has different effects under different scenes, this thesis introduces a general method for obtaining video summary.2. In this thesis, the traditional shot boundary detection algorithm is studied; and by introducing the concept of information entropy, the method of shot segmentation based on mutual information entropy is analyzed. According to the characteristics of information entropy, this thesis proposes a method to obtain the video key frame, with which the information distribution of each video frame can be extracted. Finally, the detailed steps of obtaining the key frames of the video are discussed.3.The proposed method is implemented in Windows 7, and the application program interface used in the acquisition of video frames and video stream is provided by Microsoft SDK9.0 Direct X development kit; at the same time, algorithm library provided by Open CV is chosen. Finally, using the C++ Visual 6 as the development tool, a interactive interface is implemented.The importance of video abstract is apparent and the method of obtaining the video summary is a hot spot in current research. This thesis focus on obtaining video summary. In our work, shot boundary detection and key frame extraction is the key to the video abstract system in the process of acquiring video abstract. We hope the work can be helpful to the further study.