A Video Steganalysis Method Based on Window Optimal Calibration

Abstract

The present invention discloses a window optimal calibration-based video steganalysis method. According to the window optimal calibration-based video steganalysis method of the invention, based on detection of a video motion vector steganography method, the objective of the invention is to solve the problem of low universality of a traditional difference value-based motion vector calibration method, firstly, a motion vector is calibrated by using the optimality of a motion vector error and a predicted error in a local window region; and secondly, corresponding steganalysis features are extracted from pre-calibration video frames and post-calibration video frames, and final steganalysis features are constructed by using Cartesian form to replace a difference value form. Based on the combination of the optimality of the motion vector error and the predicted vector error in the window and the Cartesian calibration form, universal and accurate steganalysis of motion vector steganography is realized. Since the calibration process of the motion vector is not limited by video coding standards and specific video coding parameters, the window optimal calibration-based video steganalysis method has high universality in application scope.

Description

technical field [0001] The invention belongs to the technical field of multimedia security and digital media processing, in particular to a steganalysis method 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 popularization of video capture equipment, digital video has become an easy-to-obtain hidden carrier; the video carrier is large in size, which can provide enough hiding space for secret information; the composition of digital video is rich and diverse, which is conducive to the design of diverse steganography method. In recent years, digital video-based steganographic techniques and tools have gradually increased, which poses a severe...

AI Technical Summary

Problems solved by technology

[0016] In the above motion vector calibration method, recompression calibration needs to use the same encoding parameters as the original video, otherwise steganalysis will have a large error, but the encoding parameters of the video are often difficult to obtain completely at the decoding end

Polynomial kernel regression calibration relies on adjacent motion vectors, but literature [6] only considers the special case of FBS and all blocks are inter-frame macroblocks (such as figure 2 As shown in a), this calibration method is not suitable for the situation where the adjacent macroblocks are inconsistently divided and the situation where the inter-frame macroblocks are discontinuous (such as figure 2 b and figure 2 c)

In addition, the existing motion vector calibration methods all use the difference form (document [2]) to construct features, which may lose some useful statistical information; and the calibration of the Cartesian form (document [3]) can be more comprehensive The retention statistic of , with more flexibility, is currently understudied to calibrate motion vectors

Images