視頻隱藏分析關(guān)鍵技術(shù)研究
發(fā)布時(shí)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(shù)引起人們的關(guān)注。為了有效檢測(cè)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?jī)r(jià)值和應(yīng)用前景。本文研究以數(shù)字視頻為載體的隱藏分析技術(shù),針對(duì)多種作用域的隱藏算法提出了相應(yīng)的檢測(cè)算法,.并將領(lǐng)域適應(yīng)與通用隱藏檢測(cè)相結(jié)合,分析其在解決載體分布失配問(wèn)題上的適用性。本文主要工作與創(chuàng)新包括以下6個(gè)方面：1、在壓縮視頻空域隱藏分析方面,提出將時(shí)域相關(guān)性和空域相關(guān)性相結(jié)合的SPEAM統(tǒng)計(jì)特征,用于檢測(cè)壓縮視頻的空域擴(kuò)頻隱藏。在非壓縮視頻的隱藏檢測(cè)性能方面,該特征與只考慮空域相關(guān)性的圖像隱藏分析特征相當(dāng),且優(yōu)于現(xiàn)有視頻隱藏分析特征；但在壓縮視頻的檢測(cè)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(yùn)算復(fù)雜度。2、在壓縮域運(yùn)動(dòng)矢量隱藏分析方面,針對(duì)現(xiàn)有運(yùn)動(dòng)矢量隱藏檢測(cè)方法或者檢測(cè)性能差且無(wú)法適應(yīng)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(diǎn),提出基于運(yùn)動(dòng)矢量局部最優(yōu)假設(shè)的AoSO特征。該特征在檢測(cè)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(fù)雜視頻編碼條件下均表現(xiàn)出較好的檢測(cè)性能。3、在H.264幀內(nèi)預(yù)測(cè)模式隱藏分析方面,結(jié)合H264視頻壓縮編碼標(biāo)準(zhǔn),提出以SATD作為編碼代價(jià)的幀內(nèi)預(yù)測(cè)模式校準(zhǔn)機(jī)制及其相應(yīng)的CIPM隱藏檢測(cè)特征。該特征在隱藏檢測(cè)性能上明顯優(yōu)于現(xiàn)有特征,且計(jì)算復(fù)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(duì)其中的隱藏算法失配問(wèn)題,對(duì)領(lǐng)域適應(yīng)中的特征遷移算法——TCA算法開(kāi)展研究,并將TCA算法用于解決隱藏算法失配條件下的運(yùn)動(dòng)矢量隱藏檢測(cè)問(wèn)題。最后通過(guò)實(shí)驗(yàn)驗(yàn)證其可能帶來(lái)失配條件下隱藏檢測(cè)性能的提升,并結(jié)合實(shí)驗(yàn)中遇到的問(wèn)題,對(duì)TCA等領(lǐng)域適應(yīng)算法的適用性進(jìn)行了探討,并指出領(lǐng)域適應(yīng)算法在隱藏檢測(cè)應(yīng)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(duì)現(xiàn)有隱藏分析方法僅考慮隱藏信息的檢測(cè)且檢測(cè)性能不理想的問(wèn)題,對(duì)MSU StegoVideo隱藏的三個(gè)指紋特征進(jìn)行了分析,包括特殊的顏色空間轉(zhuǎn)換公式、棋盤狀掩膜圖案以及嵌入強(qiáng)度控制機(jī)制,提出基于MSU StegoVideo指紋的非壓縮視頻隱藏檢測(cè)與隱藏參數(shù)估計(jì)方法——MP算法。該算法在檢測(cè)性能上遠(yuǎn)優(yōu)于已有算法,且可對(duì)隱藏參數(shù)"noise level"以及隱藏信息序列進(jìn)行初步的估計(jì)。6、在格式隱藏軟件OpenPuff的隱藏分析方面,針對(duì)該軟件支持的MPEG、VOB、 FLV、MP4、3GP等格式視頻,結(jié)合相應(yīng)的視頻容器標(biāo)準(zhǔn),通過(guò)分析隱藏操作引入的文件差異,提出相應(yīng)的隱藏分析方法,能夠?qū)崿F(xiàn)高精度的盲檢測(cè)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(xué)位授予單位】：解放軍信息工程大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2014
【分類號(hào)】：TN919.81
本文編號(hào)：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(xué)位授予單位】：解放軍信息工程大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2014
【分類號(hào)】：TN919.81
【參考文獻(xiàn)】
相關(guān)期刊論文 前10條
1 牛小鵬;李清寶;王煒;;一種基于擴(kuò)頻編碼的可靠網(wǎng)絡(luò)隱蔽信道設(shè)計(jì)方法[J];電子與信息學(xué)報(bào);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(shí)空冗余統(tǒng)計(jì)不可見(jiàn)性的視頻隱寫分析方法[J];計(jì)算機(jī)應(yīng)用研究;2012年10期
4 孫建芳;湯光明;;針對(duì)音頻抖動(dòng)調(diào)制QIM隱寫的盲檢測(cè)[J];計(jì)算機(jī)工程與應(yīng)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(xué)報(bào);2010年09期
6 陳嘉勇;祝躍飛;張衛(wèi)明;劉九芬;;對(duì)隨機(jī)LSB隱寫術(shù)的選擇密鑰提取攻擊[J];通信學(xué)報(bào);2010年05期
7 馬麗紅;謝中華;劉小燕;;視頻數(shù)據(jù)隱藏的差分能量調(diào)制新算法[J];中國(guó)圖象圖形學(xué)報(bào);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(jìn)展[J];計(jì)算機(jī)學(xué)報(bào);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(wǎng)絡(luò)信息隱藏系統(tǒng)研究[J];南京理工大學(xué)學(xué)報(bào)(自然科學(xué)版);2009年03期
10 何柳;韓偉力;;針對(duì)QIM信息隱藏的高效盲檢測(cè)算法[J];計(jì)算機(jī)工程;2008年02期
，本文編號(hào)：2342463
本文鏈接：http://sikaile.net/kejilunwen/wltx/2342463.html
最近更新
教材專著
