Video Privacy Protection FrFT Based Research

Today, the Internet openness, the resources sharing and the ease of acquiring information have brought great convenience to people’s lives. At the same time, privacy violations by using the internet are more likely to occur than before. Some organizations or individuals unscrupulously use a variety of software to collect and process personal information and then spread them on the internet, so as to achieve evil purpose or get illegitimate benefits. The illegal distribution of the personal portrait video which contains a lot of valuable information on the network is an extremely common approach. Privacy protection has become an urgent social problem which needs to be solved. At present, in addition to laws to protect people’s privacy, privacy protection technology has become a hot research topic. It has very important value both in theory and in practice.The thesis presents a method to encrypt the video based on fractional Fourier transform in order to protect the personal privacy. Firstly, detect the faces in the video by the Adaboost face detection algorithm on the OpenCV platform, and then encrypt the detected faces by using the fractional Fourier transform. The main works performed in the thesis are:1. Train the positive samples and negative samples by Adaboost algorithm to get the face classifier. The trained samples contain a total of1600positive samples which all the size is20x20, selected from the MIT face database and ORL face database and also contain a total of4381negative samples of the same size with the positive samples.2. Achieve the interlace between OpenCV and Matlab by setting the OpenCV and the Matlab environment respectively and encrypted the detected faces in the video by calling the fractional Fourier transform dynamic link library that was generated by the m function of Matlab on the OpenCV platform.The experiments show that the classifier trained by Adaboost algorithm can detect the faces in the video better and the detection rate can reach90.6%. All the detected faces also can be encrypted. After the encryption of the faces in the video through the fractional Fourier transform, only those who are authorized to get the right key can decrypt the faces correctly, otherwise people can’t see the faces. Thus it protects the portrait from abuse and achieves the goal of protecting personal privacy.