被動盲目式影像及視訊偽造偵測之研究

Passive-Blind Image and Video Forgery Detection

林承賢(Cheng-Shian Lin)

指導教授 ： 蔡志忠

國立中正大學/工學院/資訊工程學系/博士(2015年)

摘要

被動盲目式影像及視訊偽造偵測可用來偵測影像/視訊中非法篡改的區域而無需任何先前資訊，其可應用於多種不同領域中以提供必要依據，諸如從主流媒體，新聞和科學出版物，到醫療成像，刑事調查和監視系統，僅舉幾例。
在本論文中，首先，我們提出基於區塊大小一致性的被動式偽造偵測於JPEG影像方法。由於現有檢測方法對影像內容是非常敏感，且受到高密集的邊緣和紋理的干擾而造成較高的誤檢率。為克服這些問題，我們首先提出了一個增強式交叉差分濾波器，加強區塊效應和減少邊緣和紋理的干擾，並結合隨機抽樣，投票和最大似然方法改進估計區塊大小的精確度。對於區塊大小的估計，我們開發了兩個不同的隨機抽樣策略: JPEG影像的原始區塊大小估計，和區域性的區塊大小的一致性分析。最後，我們進行精鍊處理，以消除錯誤偵測，並填補未檢測到篡改區塊。所提方法能有效率偵測JPEG影像遭惡意篡改之區域。
第二，我們提出基於空間與時間上的相關性分析以偵測及定位視訊篡改區域。由於現有檢測方法只僅對畫面層級作檢測而無法定位篡改區域，亦或是定位的篡改區域具有高誤檢率。因此，在此論文中，我們探討在視訊中物件移除後常見的兩種區域性修復方法:時空複製及貼上和基於範例之紋理合成技術，並提出了一種基於時空相關分析檢測及定位篡改區域方法。我們的方法可以處理相機移動的影響和多物件移除。實驗結果顯示我們的方法是可效地檢測及定位經時空複製及貼上和紋理合成的篡改區域。
第三部分，我們提出基於異常DCT係數檢測的被動式偽造偵測於雙重壓縮JPEG影像方法。由於現有檢測方法只僅對影像層級的雙重壓縮JPEG影像作檢測而無法定位篡改區域，亦或是定位的篡改區域具有高誤檢率。因此，在此論文中，我們探討在篡改影像後常見的兩種影像儲存格式: 未壓縮 和JPEG壓縮格式，並提出了一種基於DCT係數分析檢測及定位篡改區域方法。我們的方法可以處理JPEG影像篡改後重新儲存於不同影像品質的JPEG影像壓縮格式。實驗結果顯示我們的方法是可效地檢測及定位經由複製及貼上，影像修補及影像合成等篡改方法之偽造區域。

關鍵字

被動盲目式 ； 偽造偵測

並列摘要

Passive image/video forgery detection aims to detect the traces of tamping without prior information, and it has been widely used to provide essential evidences in many diverse areas, ranging from mainstream media, journalism and scientific publication, to medical imaging, criminal investigations and surveillance systems, to name a few.
First, passive forgery detection for JPEG compressed image based on block size estimation and consistency analysis is proposed. Our investigation shows that current approaches for detection and localization of tampered areas are very sensitive to image contents, and suffer from high false detection rates for localization of tampered areas for images with intensive edges and textures. For overcoming these problems, we first propose an enhanced cross difference filter to strengthen block artifacts and reduce interference from edges and textures, and then integrate techniques from random sampling, voting and maximum likelihood method to improve the accuracy of block size estimation. We develop two different random sampling strategies for block size estimation: one for estimation of the primary JPEG block size, and the other for consistency analysis of local block sizes. We finally perform a refinement process to eliminate false detections and fill in undetected tampered blocks.
Second, passive approach for effective detection and localization of region-level video forgery with spatio-temporal coherence analysis is proposed. Most of current passive approaches either work only for frame-level detection and cannot localize region-level forgery, or suffer from high false detection rates for localization of tampered regions. In this thesis, we investigate two common region-level inpainting methods for object removal, temporal copy-and-paste and exemplar-based texture synthesis, and propose a new approach based on spatio-temporal coherence analysis for detection and localization of tampered regions. Our approach can handle camera motion and multiple object removal. Experiments show that our approach outperforms previous approaches, and can effectively detect and localize regions tampered by temporal copy-and-paste and texture synthesis.
Finally, passive forgery detection for double compressed JPEG images using DCT coefficient analysis is proposed. Most current passive approaches either work only for image-level double JPEG compression detection and cannot localize region-level forgery, or suffer from high false detection rates in localizing altered regions. In this thesis, we investigate two common image formats for saving tampered image, uncompressed and JPEG compressed format, and proposed an effective approach based on DCT coefficient analysis for the detection and localization of altered regions from JPEG compressed images. Our approach can handle the tampered JPEG image resaved in JPEG compressed format with different quality factors. Experiments with various tampering methods such as copy-and-paste, image completion and composite tampering, show that the proposed approach is able to effectively detect and localize altered areas.

並列關鍵字

Forgery Detection ； Passive-Blind

參考文獻

[1] M. Barni, A. Costanzo, and L. Sabatini, “Identification of cut & paste pampering by means of double-JPEG detection and image segmentation,” in Proc. of IEEE Int. Conf. on Circuits and Systems, 2010, pp. 1687-1690.

[2] S. Bayram, H.T. Sencar, and N. Memon, “Source camera identification based on CFA interpolation,” in Proc. of IEEE Int. Conf. on Image Processing, 2005, pp. 69-72.

[3] A. Criminisi, P. P´erez, and K. Toyama, “Region filling and object removal by exemplar-based image inpainting,” IEEE Trans. on Image Processing, vol. 13, no. 9, pp. 1200-1212, 2004.

[4] W. Chen, Y.Q. Shi, and W. Su, “Image splicing detection using 2-D phase congruency and statistical moments of characteristic function,” in Proc. of SPIE Int. Conf. on Security, Steganography, and Watermarking of Multimedia Contents, 2007, pp. 65050.

[5] Y.L. Chen, and C.T. Hsu, “Image tampering detection by blocking periodicity analysis in JPEG compressed images,” in Proc. of IEEE Int. Conf. on Multimedia Signal Process., 2008, pp. 803-808.