【摘要】：已有的大多數(shù)H.264加密算法只考慮了視頻編碼的特性,而忽略了人眼才是處理視頻信號的接收端.提出一種基于人眼視覺感知特性的運動注意力的視頻熵編碼加密算法.將所有宏塊根據(jù)視覺感知注意力模型分成I幀宏塊、中心區(qū)域宏塊、注意力較大宏塊和注意力較小宏塊.采用混沌系統(tǒng)產(chǎn)生加密密鑰,然后對四類宏塊采用不同的加密手段,先對I幀宏塊DCT系數(shù)翻轉然后進行熵編碼加密,然后引入標示宏塊在人眼中注意力強弱的視覺注意力權重分布w:當wij=1時,宏塊處于中心區(qū)域,對MV值、DCT系數(shù)的熵編碼過程加密;當wij≠1時,則計算宏塊運動量的加權,并與閾值T比較.注意力較大宏塊進行熵編碼加密,注意力較小宏塊則不做加密操作.本文算法將感知模型引入加密算法中,在同一幀內實現(xiàn)分類的加密手段.實驗表明,本算法加密后數(shù)據(jù)量僅增加了10%-15%,加密所占用額外時間較小,基本對信噪比SNR沒有造成影響,從直方圖和混淆后的圖形分析,加密的效果顯著.
[Abstract]:Most existing H. 264 encryption algorithms only consider the characteristics of video coding and ignore that the human eye is the receiver of video signal processing. An entropy-coded video encryption algorithm based on human visual perception is proposed. According to the visual perceptual attention model, all macroblocks are divided into I frame macroblocks, central area macroblocks, attention larger macroblocks and small attention macroblocks. Chaotic system is used to generate encryption keys, and then four kinds of macroblocks are encrypted by different encryption methods. First, the DCT coefficients of frame I macroblocks are flipped and then entropy coded encryption is carried out. Then the visual attention weight distribution w is introduced, which indicates the attention intensity of macroblocks in human eyes. When wij=1, the macroblocks are in the central region, and the entropy coding process of MV and DCT coefficients is encrypted. When wij 鈮，

本文編號：2440355