近年來，浮水印技術開始加入了特徵搜尋的概念，以求更強的robust能力。而近年廣受歡迎的特徵搜尋方法即為Scale Invariant Feature Transform(SIFT)，SIFT 所找出的特徵點在抵抗旋轉、縮放等破壞時，擁有高度的不變性。本篇提出了一個高capacity和高擷取品質的robust watermark方法。將watermark嵌入至SIFT所搜尋出的特徵區域，搭配NonSubsampled Contourlet Tansform(NSCT)和加入tree split、投票、旋轉搜尋和形態學的概念。我們所提出的方法在比較以往的浮水印技術中，擁有更高的capacity和更有效的抵抗影像處理破壞。

Geometry invariant features have been employed in watermarking algorithms to enhance robustness. The most widely used feature is Scale Invariant Feature Transform (SIFT) which is able to resist against various image distortion, such as rotation, scaling and shifting. In this work, the input image is first decomposed into the high-pass and low-pass subband by Non-Subsampled Contourlet Tansform (NSCT). The binary watermark image is then embedded through SIFT features extracted from the low-pass subband. Experimental results have shown that the proposed method can efficiently resist various image attacks while maintain high image fidelity.

[1] L. Li, J. Qian, and J.-S. Pan, "High capacity watermark embedding based on
local invariant features," IEEE Int. Conf. on Multimedia and Expo, pp. 1311-1314, Jul. 2010.
[2] G. C. Langelaar and R. L. Lagendijk, "Optimal di erential energy watermark-
ing of dct encoded images and video," IEEE Trans. on Image Processing, vol. 10, pp. 148-158, Jun. 2001.
[3] T. K. Das, S. Maitra, and J. Mitra, "Cryptanalysis of optimal di erential energy watermarking(dew) and a modi ed robust scheme," IEEE Trans. on Image
Processing, vol. 53, pp. 768-775, Feb. 2005.
[4] J. C. Patra, J. E. Phua, and D. Rajan, "Dct domain watermarking scheme
using chinese remainder theorem for image authentication," IEEE Int. Conf.
on Multimedia and Expo, pp. 111-116, Jul. 2010.
[5] H. Ling, Z. Lu, and F. Zou, "Improved di erential energy watermarking (idew)
algorithm for dct encoded image and video," IEEE Int. Conf. on Signal Processing, vol. 3, pp. 2326-2329, Sep. 2004.
[6] Z. Wei and K. N.Ngan, "Spatial just noticeable distortion pro le for image
in dct domain," IEEE Int. Conf. on Multimedia and Expo, pp. 925-928, Apr. 2008.
[7] Y. Niu, J. Liu, S. Krishnan, and Q. Zhang, "Combined just noticeable di erence model guided image watermarking," IEEE Int. Conf. on Multimedia and Expo, pp. 1679-1684, Jul. 2010.
[8] S. Ababneh, R. Ansari, and A. Khokhar, "Compressed-image authentication
using compensated signature embedding," IEEE Int. Conf. on Electro/Infor-
mation Tech., pp. 476-481, May 2007.
[9] S. Ababneh, A. Khokhar, and R. Ansari, "Compensated signature embedding
based multimedia content authentication system," IEEE Int. Conf. on Image
Processing, pp. I-393-I-396, Sep. 2007.
[10] S. Ababneh, R. Ansari, and A. Khokhar, "Improved image authentication us-
ing closed-form compensation and spread-spectrum watermarking," IEEE Int.
Conf. on Acoustics Speech and Signal Processing, pp. 1781-1784, Mar. 2008.
[11] H. Yuan and X.-P. Zhang, "A multiscale fragile watermark based on the gaus-
sian mixture model in the wavelet domain," IEEE Int. Conf. on Acoustics Speech and Signal Processing, pp. 413-416, May 2004.
[12] H. Yuan and X.-P. Zhang, "A secret key based multiscale fragile watermark in the wavelet domain," IEEE Int. Conf. on Multimedia and Expo, pp. 1333-1336,
Jul. 2006.
[13] X. Wang, J. Wang, and H. Peng, "A semi-fragile image watermarking resisting
to jpeg compression," IEEE Int. Conf. on Management of e-Commerce and
e-Government, pp. 498-502, Sep. 2009.
[14] D. G. Lowe, "Distinctive image features from scale-invariant keypoints," International Journal of Computer Vision, vol. 60, no. 2, pp. 91-110, 2004.
[15] A. L. Cunha, J. Zhou, and M. N. Do, "The nonsubsampled contourlet trans-
form: Theory, design, and application," IEEE Trans. on Image Processing,
vol. 15, pp. 3089-3101, Oct. 2006.
[16] Canny and John, "A computational approach to edge detection," IEEE Trans.
on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no. 6, pp. 679-698,
1986.
[17] http://sipi.usc.edu/database.