Research On The Pornographic Video Detection Algorithm Based On Optic Flow Features In Shots

With the arrival of the networked and informatization era, the Network is walking into people's life with its powerful convenient information transmission. People are often disturbed with some negative information such as violence, sex and fraud information, when they are acquiring knowledge and entertainment, and communicating with others and doing business on the internet. In order to detect the negative information effectively and block its transmission way quickly, the study on the technology of identification and detection base on the negative network information has caught people's comprehensive attention. The negative information is often spreading on the internet in the form of the text, images, video and music and so on. Aimed at the over-flooding of pornographic video, a pornographic video detection algorithm base on the property of optical flow field in a shot is proposed.The algorithm is divided into two parts in the main. We divide the video into a lot of shots with the characteristics of the color and edge of the frames first. We divide the video into a lot of shots with the characteristics of the color and edge of the frames first. Then extract the characteristic of the optical flow of the video sequence. Finally, we get the judgment according to the detection of the human's reciprocating motion in the pornographic video. The main work of this paper is as follows:1. According to the global color's characteristic of sequence frame of the video, detect the abrupt shot by setting the adaptive threshold. For the flash interference in the video, we will eliminate it with the brightness information of the edge image, so as to improve the accuracy of the segmentation lens.2. Divide the video into several small video on the base of the determination of the abrupt shot's boundary. Then detect the gradual shot in these video segmentations orderly. Constructing the 1 gradient shot detector to determine the segmentation of boundary using the image's edge character according to the different classification of the gradual shot, so as to optimize the result of the shot segmentation greatly.3. On the basis of calculating the optical flow field, we extract the characteristics points of the moving target, and classify optical flow direction of these points and get the statistics of them. Then we establish the optical flow direction's statistical histogram. According to the changes in the direction of moving target and the changes in the frequency of it, we can determine there is pornographic information or not.