Hevc double-compressed video detection method based on space-time complexity measure and local prediction residual distribution

Abstract

The invention discloses a HEVC double-compression video detection method based on spatio-temporal complexity measurement and local prediction residual distribution, which comprises the following steps: decompressing the input video into a frame sequence, calculating the key points of each frame and comparing two adjacent frames Carry out key point matching; for the key points of each video frame, calculate the space-time complexity measure of the local area centered on the key point; sort the key points according to the space-time complexity measure, and eliminate the key points with spatial aggregation relationship; The remaining key points of each frame are processed to construct the feature sequence of the input video; finally, the feature sequence is divided into non-overlapping sub-sequences, and the sub-sequences are input into the trained multi-layer perceptron classifier for subsequent processing, so as to Whether the obtained input video output score is greater than the threshold is used as the judgment of HEVC dual compression video. The detection method of the present invention can have good detection robustness to various video contents and encoding parameter settings.

Description

technical field [0001] The present invention relates to the technical field of video re-compression detection methods, in particular to a HEVC double-compression video detection method based on time-space complexity measurement and local prediction residual distribution. Background technique [0002] With the rapid development of information technology, multimedia information carriers such as digital video have been widely used in people's daily life, including fields such as news, medical treatment and education. However, more and more technologies are mature and simple video editing software is available so that users can easily modify video content without leaving traces perceptible to the naked eye. If the maliciously tampered digital video is used illegally, it will cause huge economic losses and serious security risks to the society, for example, the tampered video is used to forge judicial electronic evidence. Generally speaking, generating a tampered video requires ...

AI Technical Summary

Problems solved by technology

The main defect of this technical solution is that only the information of the first P frame in the GOP structure is considered, and the detection performance of the heavily compressed video with strong content changes is degraded

The main defect of this technical solution is that the input video is required to be compressed with a fixed GOP structure, and it cannot provide reliable detection for double-compressed video spliced ​​with an adaptive GOP structure or different GOP structures.

Images