以SIFT與區塊擴張技術進行複製移動竄改區域偵測

Detecting Copy-Move Forgery Regions through SIFT and Region Growing Strategies

盧威宇(Wei-Yu Lu)

指導教授 ： 陳建彰

淡江大學/工學院/資訊工程學系碩士班/碩士(2017年)

https://doi.org/10.6846/TKU.2017.00363

摘要

數位相機及智慧型手機已成為生活中不可缺少的部分，因此我們隨處可見數位影像，但由於數位化資料擁有容易修改的特性，且影像處理軟體功能日益強大的狀況下，往往可以記錄真實事物的特性面臨極大的考驗。
本論文以區塊擴張及SIFT特徵點之使用來探討區塊比對的複製移動竄改偵測技術，先使用SIFT關鍵點內的資訊作為特徵點的提取及角度的計算，再利用SIFT的關鍵點位置，對影像以9 9大小進行擷取，並進行特徵提取進行，若在於門檻值之內，就以同心矩形的方式依序擴大，再來進行小範圍的區域擴張，使得偵測的圖形可以更加完整。經實驗結果本論文研究利用多種不同的特徵值集合，可以有效降低偵測的時間，及改善前人因竄改區域變化無法偵測的問題。

關鍵字

複製移動竄改 ； 不變動量 ； 尺度不變特徵轉換

並列摘要

Since digital cameras and smart phones have become an indispensable part of life, a lot of digital images are widely used. However, digital image has the property of easy to modify through image editing software, the ability to find real image is a great challenge.
This paper presents an algorithm to detect copy-move image regions by SIFT keypoints and region growing technique. First, the SIFT keypoints denotes important image’s feature points. By using scale similarity comparison technique, the similar starting 99 pair blocks are acquired. The region growing technique is adopted to generate the copy-move regions. Experimental results show that the SIFT keypoints are useful to detect starting copy-move blocks and further image-growing technique can detect copy-move regions effectively.

並列關鍵字

Copy-Move Forgery ； Invariant Moment ； SIFT

參考文獻

[1] X. Bi, C. Pun and X. Yuan, “Multi-level dense descriptor and hierarchical feature matching for Copy–Move forgery detection,” Information Sciences, 345, pp. 226-242. 2016.

[2] S. Bayram, H. Taha Sencar and N. Memon, “An efficient and robust method for detecting copy-move forgery,” IEEE International Conference on Acoustics, Speech and Signal Processing, 2009.

[3] C. C. Chen, H. Wang, and C. S. Lin. “An efficiency enhanced cluster expanding block algorithm for copy-move forgery detection,” Multimedia Tools and Applications, pp. 1-20, 2016.

[5] R. Davarzani, K. Yaghmaie, S. Mozaffari and M. Tapak, “Copy-move forgery detection using multiresolution local binary patterns,” Forensic Science International, 231(1–3), pp. 61-72, 2013.

[8] M. K. Hu, “Visual pattern recognition by moment invariants,” IRE Transactions on Information Theory 8(2), pp. 179-187, 1962.