視頻隱藏分析關鍵技術研究
發(fā)布時間：2018-11-19 13:39
【摘要】：作為大容量信息隱秘通信的有效手段,視頻信息隱藏技術引起人們的關注。為了有效檢測以視頻為載體的有害隱秘通信行為,視頻隱藏分析具有重要的研究價值和應用前景。本文研究以數(shù)字視頻為載體的隱藏分析技術,針對多種作用域的隱藏算法提出了相應的檢測算法,.并將領域適應與通用隱藏檢測相結合,分析其在解決載體分布失配問題上的適用性。本文主要工作與創(chuàng)新包括以下6個方面：1、在壓縮視頻空域隱藏分析方面,提出將時域相關性和空域相關性相結合的SPEAM統(tǒng)計特征,用于檢測壓縮視頻的空域擴頻隱藏。在非壓縮視頻的隱藏檢測性能方面,該特征與只考慮空域相關性的圖像隱藏分析特征相當,且優(yōu)于現(xiàn)有視頻隱藏分析特征；但在壓縮視頻的檢測性能方面,該特征優(yōu)于已有的視頻隱藏分析特征和圖像隱藏分析特征,且具有較低的運算復雜度。2、在壓縮域運動矢量隱藏分析方面,針對現(xiàn)有運動矢量隱藏檢測方法或者檢測性能差且無法適應多種視頻編碼格式,或者因重壓縮參數(shù)與原始編碼參數(shù)不同而失效的缺點,提出基于運動矢量局部最優(yōu)假設的AoSO特征。該特征在檢測性能上明顯優(yōu)于現(xiàn)有的兩類特征,且適用范圍較廣,在多種復雜視頻編碼條件下均表現(xiàn)出較好的檢測性能。3、在H.264幀內預測模式隱藏分析方面,結合H264視頻壓縮編碼標準,提出以SATD作為編碼代價的幀內預測模式校準機制及其相應的CIPM隱藏檢測特征。該特征在隱藏檢測性能上明顯優(yōu)于現(xiàn)有特征,且計算復雜度比較低。4、在改善載體分布失配條件下的隱藏分析算法性能方面,針對其中的隱藏算法失配問題,對領域適應中的特征遷移算法——TCA算法開展研究,并將TCA算法用于解決隱藏算法失配條件下的運動矢量隱藏檢測問題。最后通過實驗驗證其可能帶來失配條件下隱藏檢測性能的提升,并結合實驗中遇到的問題,對TCA等領域適應算法的適用性進行了探討,并指出領域適應算法在隱藏檢測應用中的局限性。5、在視頻空域隱藏軟件MSU StegoVideo的隱藏分析方面,.針對現(xiàn)有隱藏分析方法僅考慮隱藏信息的檢測且檢測性能不理想的問題,對MSU StegoVideo隱藏的三個指紋特征進行了分析,包括特殊的顏色空間轉換公式、棋盤狀掩膜圖案以及嵌入強度控制機制,提出基于MSU StegoVideo指紋的非壓縮視頻隱藏檢測與隱藏參數(shù)估計方法——MP算法。該算法在檢測性能上遠優(yōu)于已有算法,且可對隱藏參數(shù)"noise level"以及隱藏信息序列進行初步的估計。6、在格式隱藏軟件OpenPuff的隱藏分析方面,針對該軟件支持的MPEG、VOB、 FLV、MP4、3GP等格式視頻,結合相應的視頻容器標準,通過分析隱藏操作引入的文件差異,提出相應的隱藏分析方法,能夠實現(xiàn)高精度的盲檢測與部分隱藏信息的提取。
[Abstract]:As an effective means of large capacity information secret communication, video information hiding technology has attracted people's attention. In order to detect the harmful hidden communication behavior with video as the carrier, video hiding analysis has important research value and application prospect. In this paper, the hiding analysis technology based on digital video is studied, and the corresponding detection algorithm is proposed for various scope hiding algorithms. Combining domain adaptation with general hidden detection, its applicability in solving mismatch of carrier distribution is analyzed. The main work and innovations of this paper are as follows: 1. In the analysis of compressed video spatial domain hiding, a SPEAM statistical feature combining temporal correlation with spatial domain correlation is proposed to detect spatial spread spectrum hiding of compressed video. In the aspect of hiding detection performance of uncompressed video, this feature is comparable to that of image hiding analysis which only considers spatial correlation, and is superior to the existing video hiding analysis feature. However, in the detection performance of compressed video, this feature is superior to the existing features of video hiding analysis and image hiding analysis, and has lower computational complexity. 2. Aiming at the shortcomings of the existing motion vector hiding detection methods, such as poor detection performance and being unable to adapt to various video coding formats, or because the recompression parameters are different from the original coding parameters, The AoSO feature based on the local optimal assumption of motion vector is proposed. This feature is obviously superior to the existing two kinds of features in detection performance, and has a wide range of applications. It has good detection performance under various complex video coding conditions. 3. In H.264 intra prediction mode hiding analysis, Based on the H264 video compression coding standard, an intra prediction mode calibration mechanism with SATD coding cost and its corresponding CIPM hidden detection features are proposed. This feature is obviously superior to the existing features in the performance of hiding detection, and its computational complexity is relatively low. 4. In order to improve the performance of the hidden analysis algorithm under the mismatch condition of carrier distribution, the proposed algorithm mismatches the hidden algorithm. The feature migration algorithm (TCA) in domain adaptation is studied, and the TCA algorithm is used to solve the problem of motion vector hiding under the condition of mismatch. Finally, it is verified by experiments that it may bring about the improvement of the performance of hidden detection under mismatch conditions, and the applicability of adaptive algorithms in TCA and other fields is discussed in combination with the problems encountered in the experiments. It also points out the limitation of domain adaptation algorithm in the application of hiding detection. 5. In the aspect of hiding analysis of video spatial hiding software MSU StegoVideo. Aiming at the problem that the existing hidden analysis methods only consider the detection of hidden information and the detection performance is not satisfactory, three fingerprint features hidden by MSU StegoVideo are analyzed, including the special color space conversion formula. The chessboard mask pattern and embedding intensity control mechanism are discussed. An uncompressed video hiding detection and parameter estimation method based on MSU StegoVideo fingerprint is proposed, which is called MP algorithm. The algorithm is much better than the existing algorithm in detection performance, and can estimate the hidden parameter "noise level" and the hidden information sequence. 6. In the aspect of hiding analysis of format hiding software OpenPuff, the MPEG,VOB, FLV, supported by the software can be analyzed. Based on the video container standard of MP4,3GP and the corresponding video container standard, the paper puts forward the corresponding hidden analysis method by analyzing the file difference caused by hiding operation, which can achieve high precision blind detection and partial hidden information extraction.
【學位授予單位】：解放軍信息工程大學
【學位級別】：博士
【學位授予年份】：2014
【分類號】：TN919.81
本文編號：2342463
[Abstract]:As an effective means of large capacity information secret communication, video information hiding technology has attracted people's attention. In order to detect the harmful hidden communication behavior with video as the carrier, video hiding analysis has important research value and application prospect. In this paper, the hiding analysis technology based on digital video is studied, and the corresponding detection algorithm is proposed for various scope hiding algorithms. Combining domain adaptation with general hidden detection, its applicability in solving mismatch of carrier distribution is analyzed. The main work and innovations of this paper are as follows: 1. In the analysis of compressed video spatial domain hiding, a SPEAM statistical feature combining temporal correlation with spatial domain correlation is proposed to detect spatial spread spectrum hiding of compressed video. In the aspect of hiding detection performance of uncompressed video, this feature is comparable to that of image hiding analysis which only considers spatial correlation, and is superior to the existing video hiding analysis feature. However, in the detection performance of compressed video, this feature is superior to the existing features of video hiding analysis and image hiding analysis, and has lower computational complexity. 2. Aiming at the shortcomings of the existing motion vector hiding detection methods, such as poor detection performance and being unable to adapt to various video coding formats, or because the recompression parameters are different from the original coding parameters, The AoSO feature based on the local optimal assumption of motion vector is proposed. This feature is obviously superior to the existing two kinds of features in detection performance, and has a wide range of applications. It has good detection performance under various complex video coding conditions. 3. In H.264 intra prediction mode hiding analysis, Based on the H264 video compression coding standard, an intra prediction mode calibration mechanism with SATD coding cost and its corresponding CIPM hidden detection features are proposed. This feature is obviously superior to the existing features in the performance of hiding detection, and its computational complexity is relatively low. 4. In order to improve the performance of the hidden analysis algorithm under the mismatch condition of carrier distribution, the proposed algorithm mismatches the hidden algorithm. The feature migration algorithm (TCA) in domain adaptation is studied, and the TCA algorithm is used to solve the problem of motion vector hiding under the condition of mismatch. Finally, it is verified by experiments that it may bring about the improvement of the performance of hidden detection under mismatch conditions, and the applicability of adaptive algorithms in TCA and other fields is discussed in combination with the problems encountered in the experiments. It also points out the limitation of domain adaptation algorithm in the application of hiding detection. 5. In the aspect of hiding analysis of video spatial hiding software MSU StegoVideo. Aiming at the problem that the existing hidden analysis methods only consider the detection of hidden information and the detection performance is not satisfactory, three fingerprint features hidden by MSU StegoVideo are analyzed, including the special color space conversion formula. The chessboard mask pattern and embedding intensity control mechanism are discussed. An uncompressed video hiding detection and parameter estimation method based on MSU StegoVideo fingerprint is proposed, which is called MP algorithm. The algorithm is much better than the existing algorithm in detection performance, and can estimate the hidden parameter "noise level" and the hidden information sequence. 6. In the aspect of hiding analysis of format hiding software OpenPuff, the MPEG,VOB, FLV, supported by the software can be analyzed. Based on the video container standard of MP4,3GP and the corresponding video container standard, the paper puts forward the corresponding hidden analysis method by analyzing the file difference caused by hiding operation, which can achieve high precision blind detection and partial hidden information extraction.
【學位授予單位】：解放軍信息工程大學
【學位級別】：博士
【學位授予年份】：2014
【分類號】：TN919.81
【參考文獻】
相關期刊論文 前10條
1 牛小鵬;李清寶;王煒;;一種基于擴頻編碼的可靠網絡隱蔽信道設計方法[J];電子與信息學報;2013年04期
2 ;Steganalysis of MSU Stego Video Based on Block Matching of Interframe Collusion and Motion Detection[J];Wuhan University Journal of Natural Sciences;2012年05期
3 鐘衛(wèi)東;吳俊卿;吳耕銳;楊海濱;;一種基于時空冗余統(tǒng)計不可見性的視頻隱寫分析方法[J];計算機應用研究;2012年10期
4 孫建芳;湯光明;;針對音頻抖動調制QIM隱寫的盲檢測[J];計算機工程與應用;2011年01期
5 王永吉;吳敬征;曾海濤;丁麗萍;廖曉鋒;;隱蔽信道研究[J];軟件學報;2010年09期
6 陳嘉勇;祝躍飛;張衛(wèi)明;劉九芬;;對隨機LSB隱寫術的選擇密鑰提取攻擊[J];通信學報;2010年05期
7 馬麗紅;謝中華;劉小燕;;視頻數(shù)據隱藏的差分能量調制新算法[J];中國圖象圖形學報;2010年01期
8 王朔中;張新鵬;張衛(wèi)明;;以數(shù)字圖像為載體的隱寫分析研究進展[J];計算機學報;2009年07期
9 錢玉文;王飛;孔建壽;王執(zhí)銓;;基于多智能體的網絡信息隱藏系統(tǒng)研究[J];南京理工大學學報(自然科學版);2009年03期
10 何柳;韓偉力;;針對QIM信息隱藏的高效盲檢測算法[J];計算機工程;2008年02期
，本文編號：2342463
本文鏈接：http://sikaile.net/kejilunwen/wltx/2342463.html
最近更新
教材專著
