Research And Application On The Key Technologies Of Sensitive Video Recognition

The rapid development of the Internet brings lots of convenience to people’s lives, but also providing more ways for the cyber criminals. So to guide the teen-agers surfing the Internet healthy and accord to the law to crack down on pornographic information dissemination has become one of the essential work, of the Internet. Web filtering is one of the effective means to stop the bad information, and bad information filtering effectively is of great significance to the Internet environment greening. In this paper, we work on the network video files and starting from the video content. We use the basic method of target recognition and detection to help us to research of the key technology of sensitivity video detection. The research contents include as follows.Firstly, research the methods of video and image processing. Learn the content of videos by the key-frames based on the analysis of video information. So we transform the video content into the key-frames by decoding and analysis the video file that makes the input data more effective and accurate. That also makes the whole detection system more efficiency.Secondly, research on the key technology of target recognition and detection. Through the method of target recognition and detection, machine learning and pattern recognition. Then combine with characteristics of the sensitive video to design the process and analysis the advantages and disadvantages of the key technology.Finally, research and implement of sensitive video detection. According to the characteristics of the sensitive video’s content choose color feature and texture feature to research. For the color feature, the sensitive video has a large number of nude so we can choose the skin color as the primary criterion. We also can use the skin color to find the target location and region of interest extraction. For the texture feature, because the video screen color variety and the color feature can’t be used as a necessary and sufficient condition of judging. We use the texture features to make the results more accurate.In this paper, through the research on the key technology of sensitive video, put forward a kind of sensitive video recognition method based on skin color and sensitive part detection. This method can detect the video activity and can be applied to the online detection of network environment. That will play a positive role to purify the net work environment. It can also be used for video auditing before that release. It is a progress of artificial intelligence.