Content-Based Fast Video Browsing System

for the present, Cloud Computing and Big Data are becoming hot topics in the Internet Industry; Data Mining is going on like a raging fire and video industry is particularly outstanding. Correspondingly, people are living in a time which information is under large explosion. There is a large amounts of data and this data is growing by exponential order. Therefore, it becoming more and more difficult for us to find information we are interested in. Especially in video aspects, it becomes a hot topic how can we quickly make any judgments about the video that we find or download.Thus, we research and design content based fast video browsing system. We extract a group of key frames to represent the whole video so that users can quickly learn about the video by browsing the key framesThe main works we have done are listed below:(1) we use QT, Opencv and git etc. We build a fast browsing system by adopting the simple single threshold algori thm based on color histogram, moreover, we do parallel development under both winXP OS and ubuntu10.04OS. The system has good compatibility and can deal with all the video format that ffmpeg backs.(2) In order to learn about video plot on the high levels of semantics, we bring in face detection. We make a judgment about plot development by calculating the amount of the faces in each key frame.(3) In the process of my experiment, we found that the classifier which is established by using haar-like feature and adaboost algorithm can not generate excellent performance. Face detection using the classifier is not precise enough, so that A large number of key frames we got is invalid. Therefore, In order to enhance fault tolerance of our system, we put forward a new idea, thai is set algorithms of face number. The set algorithms improve the fault tolerance of the whole system greatly when it is introduced to.