【摘要】：近幾年,以數(shù)字圖像和數(shù)字視頻為代表的多媒體技術(shù)的飛速發(fā)展以及高質(zhì)量便攜的數(shù)碼攝像設(shè)備的廣泛普及,多種多樣的數(shù)字多媒體資源給人們的生活、工作帶來了不少的方便和樂趣,于此同時,越來越多的圖像視頻編輯軟件應(yīng)運而生,然而一些不法之徒卻利用其易獲得性和易操作性的特點惡意制造虛假視頻,散發(fā)謠言,敲詐勒索,嚴重擾亂了社會秩序,給信息安全帶來了嚴峻挑戰(zhàn)。因此如何快速有效檢測數(shù)字視頻的原始性、完整性和真實性成為數(shù)字視頻安全領(lǐng)域人們?nèi)找骊P(guān)注的課題之一。數(shù)字視頻被動取證技術(shù)就是在這種情況下產(chǎn)生的。本論文首先對數(shù)字視頻被動取證技術(shù)研究現(xiàn)狀進行概述,隨后圍繞視頻內(nèi)容真實性篡改檢測中的關(guān)鍵問題展開研究和探索。內(nèi)容篡改中以對象級篡改最為具有迷惑性。對象級篡改是指以前景對象為目標,對視頻進行篡改,主要用于區(qū)別整幀篡改。本文主要針對對象級數(shù)字視頻被動取證技術(shù)在以下三個方面開展工作:(1)針對對象移除篡改操作,通過對比高頻分量差異度來對視頻對象移除篡改進行檢測。由對篡改軟件分析得出在視頻篡改時,為了達到視覺連貫性的目的,篡改軟件通常會引入相類似的修復(fù)操作,這就使得篡改區(qū)域高頻分量差異性較小。實驗結(jié)果表明該算法取得了較好的基于子集的漏警率、虛警率,分別為7.20%和3.07%。(2)在上述方法的基礎(chǔ)上,分別從時域、空域兩個角度挖掘互補特征,提出了一種針對對象移除視頻的空時域聯(lián)合檢測算法。在時域提取殘留信息時,從視頻剖面圖的角度嘗試分析篡改區(qū)域,此方法不僅達到了降維的目的,還使得篡改痕跡更加凸顯。實驗結(jié)果表明該方法相較(1)方法檢測率得到了進一步提升。(3)針對異源視頻合成篡改操作,利用光流易受拍攝環(huán)境影響的特征,通過對比合成視頻中不同前景對象的光流,來判斷前景對象是否來自同一拍攝環(huán)境。異源視頻合成篡改是指將不同視頻內(nèi)的前景對象合成到同一個視頻背景下的操作。因光流除與物體速度有關(guān)外,其大小也會受不同拍攝環(huán)境下光照強度的影響,所以排除速度因素影響后的光流值還具有表征背景環(huán)境的作用。
[Abstract]:In recent years, with the rapid development of multimedia technology, represented by digital images and video, and the widespread popularization of high quality portable digital camera equipment, various kinds of digital multimedia resources have given people's lives. Work has brought a lot of convenience and pleasure. At the same time, more and more image and video editing software has come into being. However, some lawbreakers have made malicious use of their characteristics of accessibility and maneuverability to create false videos and disseminate rumors. Extortion seriously disturbs social order and brings severe challenge to information security. Therefore, how to quickly and effectively detect the originality, integrity and authenticity of digital video has become one of the increasingly concerned topics in the field of digital video security. Digital video passive forensics technology is produced in this situation. This paper firstly summarizes the research status of digital video passive forensics, and then focuses on the key issues of video content authenticity tampering detection. Object level tampering is the most puzzling in content tampering. Object-level tampering refers to tampering video with foreground object, which is mainly used to distinguish the whole frame tampering. This paper mainly focuses on the object level digital video passive forensics technology in the following three aspects: (1) for the object removal tamper operation by comparing the high-frequency component difference to detect the video object removal tamper. From the analysis of tampering software, in order to achieve visual coherence, tampering software usually introduces similar repair operations, which makes the difference of high-frequency components in tamper region small. The experimental results show that the proposed algorithm achieves better subsets based false alarm rate (7.20%) and false alarm rate (3.07%). (2) on the basis of the above methods, the complementary features are mined from the time domain and the spatial domain, respectively. A joint space-time detection algorithm for object removal video is proposed. When the residual information is extracted in time domain, we try to analyze the tampering area from the angle of video profile. This method not only achieves the purpose of reducing the dimension, but also makes the tampering trace more prominent. The experimental results show that the detection rate of this method is further improved than that of (1) method. (3) aiming at the heterogenous video synthesis tampering operation, the optical flow of different foreground objects in the synthesized video is compared by using the characteristics of the optical flow which is easy to be affected by the shooting environment. To determine whether the foreground object is from the same shooting environment. Heterogeneous video synthesis tampering refers to the operation of synthesizing foreground objects in different video into the same video background. Because the light flow is not only related to the velocity of the object, but also affected by the intensity of light in different shooting environment, so the value of the light flow after excluding the influence of the velocity factor also has the function of representing the background environment.
【學(xué)位授予單位】：寧波大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TP391.41;D918.2

【相似文獻】

相關(guān)期刊論文 前10條

1 丁秋峰;孫國梓;;云計算環(huán)境下取證技術(shù)研究[J];信息網(wǎng)絡(luò)安全;2011年11期

2 朱翔;;電子取證技術(shù)的發(fā)展與規(guī)范[J];警察技術(shù);2006年04期

3 田志宏;姜偉;張宏莉;;一種支持犯罪重現(xiàn)的按需取證技術(shù)[J];清華大學(xué)學(xué)報(自然科學(xué)版);2014年01期

4 徐宏斌;;計算機動態(tài)取證技術(shù)研究[J];貴州警官職業(yè)學(xué)院學(xué)報;2007年01期

5 ;智能手機取證技術(shù)研究[J];警察技術(shù);2012年05期

6 劉聰穎;陳曉平;張鵬;;電子犯罪取證淺談——技術(shù)的發(fā)展及國外的相關(guān)法律[J];科技信息;2014年10期

7 劉澤;;信息化警務(wù)模式下網(wǎng)絡(luò)取證技術(shù)完善的研究[J];網(wǎng)絡(luò)安全技術(shù)與應(yīng)用;2012年01期

8 彭建新;周元建;;iOS設(shè)備取證技術(shù)研究[J];中國人民公安大學(xué)學(xué)報(自然科學(xué)版);2012年04期

9 趙振洲;喬明秋;;基于ATT-3000的動態(tài)仿真取證技術(shù)研究[J];北京政法職業(yè)學(xué)院學(xué)報;2010年04期

10 ;英國運用遠程取證技術(shù)打擊網(wǎng)絡(luò)犯罪[J];計算機與網(wǎng)絡(luò);2007年16期

相關(guān)會議論文 前10條

1 趙志巖;王任華;邵，

本文編號：2138527