Neighborhood Optimal Probability Video Steganalysis Method and System Based on Segmentation

Abstract

The invention discloses a partition-based video steganography analysis method and system having the neighbourhood optimal probability. The method comprises the following steps of: selecting one from a partition mode set according to features of a video; for one video frame, counting a set of neighbourhood optimal SADs in all macro-block partition modes, calculating steganography analysis features of one video frame, and combining the features to obtain final steganography analysis features of one video frame; and, according to this, sequentially extracting steganography analysis features of all frames in one video. Oriented on detection of a video motion vector steganography method, aiming at the problem that the traditional prediction residual-based steganography analysis features are easily interfered by quantization distortion, the partition-based video steganography analysis method and system disclosed by the invention realize universal and precise steganography analysis to the motion vector steganography method through combination of quantization of macro-block partition to quantization distortion and neighbourhood optimal probability features.

Description

technical field [0001] The invention relates to the technical field of multimedia security and digital media processing, in particular to a steganalysis technical solution for judging whether a digital video is embedded with secret information. Background technique [0002] Modern steganography is a technology that uses digital media for secret communication, and steganalysis is a reverse detection technology of steganography, whose goal is to determine whether there is secret information hidden in digital media such as images, audio, and video. With the popularity of video capture equipment and the popularity of Internet video applications, digital video has become an easy-to-obtain hidden carrier; the video carrier has a large volume and can provide enough hidden space for secret information. In recent years, digital video-based steganographic techniques and tools have gradually increased, which poses a severe challenge to digital video steganalysis. [0003] Since H.264 / ...

AI Technical Summary

Problems solved by technology

Steganalysis features based on prediction residuals can only be constructed from the decoding end of the video, so such features are inevitably disturbed by quantization distortion

In addition, the steganalysis feature based on the prediction residual only utilizes the local optimality of SAD [3][4], while the modification of the motion vector by steganographic embedding also changes the statistical law of the neighborhood optimality of SAD, At present, no research has proposed related technical solutions

Images