應用於醫學影像感興趣區域之可偵測竄改與復原的浮水印技術

A Digital Watermarking Technique for Tamper Detection and Recovery in ROI of Medical Images

莊曜陽(Yao-Yang Chuang)

指導教授 ： 繆紹綱

中原大學/工學院/電子工程研究所/碩士(2009年)

https://doi.org/10.6840/CYCU.2009.00842

摘要

隨著網際網路的快速發展，數位媒體資料的安全保護成為備受矚目的焦點，其中也包括防止醫學影像的惡意散佈和竄改，因應此種需求，數位浮水印技術隨之逐漸受到了重視。本文針對醫學影像提出一可偵測竄改與復原的數位浮水印方法，結合強健型與脆弱型浮水印的優點，並依照醫學影像之區域重要性區分，將兩種不同類型的浮水印分別嵌入其適合的區域。在影像中不感興趣區域(NROI)，浮水印使用強健型的嵌入方法，使浮水印具有抵抗攻擊的能力，可用以辨識驗證影像的來源所屬，而在影像中感興趣區域(ROI)，浮水印則使用脆弱型的嵌入方法，主要用以確認資料的完整性，另外在整張影像中，還嵌入了影像特徵資訊，可將影像被竄改區域定位出來並將其復原。
本文以一些常見的數位影像處理攻擊來測試所提之方法，並模擬醫學影像中有病徵影像與健康影像的相互竄改來實驗定位復原的能力。實驗結果顯示所提方法的NROI浮水印對攻擊具有抵抗性，且從ROI取回的浮水印是否被破壞可以得知影像是否遭受竄改，而被竄改區域利用取回的特徵資訊，亦能夠將之定位復原，因此本文所提方法可有效防止醫學影像的惡意散佈和竄改，增加其保護及安全性。

關鍵字

數位浮水印 ； 醫學影像 ； ROI ； 竄改區域定位 ； 復原

並列摘要

As the fast development of the Internet, securing digital media data has become the focus of attention. Preventing medical images from being maliciously disseminated and tampered belongs to such an issue. As a solution to this issue, digital watermarking techniques become more and more important. In this thesis, we propose a digital watermarking technique for tamper detection and recovery for medical images. The proposed method combines the advantages of robust and fragile watermarks, and according to the corresponding importance of different medical image regions, two different watermarks are embedded in separate and suitable regions. In the region of less or no interest (NROI), a robust watermark is embedded in it. This watermark is capable of resisting image attacks, and can be used to identify and verify the ownership of the image. In the region of interest (ROI), a fragile watermark is embedded in it to verify the integrality of the data. Moreover, the characteristic information of image is also embedded in the whole image. The tampered regions of the image can be located and recovered by using this information.
To test the proposed method, we perform some digital image processing attacks and tamper the medical images such that they look from healthy to sick or from sick to healthy. The experimental results show that the NROI watermark can resist the attacks well, and a tamper of the image is successfully identified from the extracted ROI watermark. And also, the tampered regions can be located and recovered by using the extracted characteristic information. Consequently, the proposed method can effectively protect the medical images from being maliciously disseminated and tempered and enhance their data security.

並列關鍵字

medical images ； ROI ； recovery ； tempered-region locating ； digital watermark

參考文獻

[4] Y. S. Lim and D. D. Feng, “Multiple Block Based Authentication Watermarking for Distribution of Medical Images,” IEEE Int. Symposium on Intelligent Multimedia, Video and Speech Processing, pp. 631-634, Oct. 2004.

[5] P. L. Lin, C. K. Hsieh, and P. W. Huang, “Hierarchical Watermarking Scheme for Image Authentication and Recovery,” IEEE Int. Conf. on Multimedia and Expo, vol. 2, pp. 963-966, June 2004.

[6] P. L. Lin, P. W. Hiang, and A. W. Peng, “A Fragile Watermarking Scheme for Image Authentication with Localization and Recovery,” IEEE 6th Int. Symposium on Multimedia Software Engineering, pp. 146-153, Dec. 2004.

[7] G. Wang and N. N. Rao, “A Fragile Watermarking Scheme for Medical Image,” IEEE-EMBS 27th Annual Int. Conf. of the Engineering in Medicine and Biology Society, pp. 3406-3409, Jan. 2006.

[9] H. Yang and A. C. Kot, “Binary Image Authentication with Tampering Localization by Embedding Cryptographic Signature and Block Identifer,” IEEE Signal Processing Letters, vol. 13, pp. 741-744, Dec. 2006.