论文部分内容阅读
针对高效视频编码(HEVC)帧内预测模式的信息隐藏算法,提出了一种隐写检测算法。基于帧内预测模式的信息隐藏算法,通过修改原始视频的帧内预测模式完成隐秘信息的嵌入。实验表明,这种操作会导致部分帧内预测单元(PU)的大小(4×4、8×8、16×16、32×32和64×64)发生改变。对隐写视频进行相同参数的重压缩后发现,不同大小的PU在数量上有向原始视频恢复的倾向,而原始视频重压缩后PU数量变化很小。定义了视频重压缩前后各尺寸PU数量及占有比例的变化率,并以此以作为特征,结合支持向量机(SVM)分类器进行视频信息隐藏检测。实验结果表明,本文算法简单、高效,而且对不同分辨率和不同量化参数的视频均有良好的检测效果。
Aiming at the information hiding algorithm of intra prediction mode of High Efficiency Video Coding (HEVC), a steganalysis algorithm is proposed. Based on the intra-prediction mode of information hiding algorithm, by modifying the original video intra-prediction mode to complete the embedding of the secret information. Experiments show that this operation will lead to some of the intra prediction unit (PU) size (4 × 4,8 × 8,16 × 16,32 × 32 and 64 × 64) change. After compressing the same parameters of the steganographic video, it is found that the PUs of different sizes have the tendency of recovering the original video in quantity, while the quantity of PUs after the original video is compressed has little change. The definition of video compression before and after the size of the number of PU and the proportion of the rate of change, and as a feature, combined with Support Vector Machine (SVM) classifier for video information hiding detection. The experimental results show that the proposed algorithm is simple and efficient, and has good detection performance for videos with different resolutions and different quantization parameters.