影像辨識的基本理念是產生一組可用視覺感知的雜湊值，所謂視覺感知的雜湊值必須要以人眼的觀點出發，亦即人眼可輕易辨識相同類型的影像，雜湊值就必須一樣或非常相近。對此，有兩個特性是需要注意的，分別是安全性與強健性，安全性要能正確偵測出影像是否遭受惡意的篡改或修飾；強健性是要能容忍一般對影像的適當修改，如旋轉、放大縮小、不同格式的轉換等等。

本論文是以Vishal Monga論文中提到的奇異值分解演算法為基礎，嘗試探討此演算法各個參數的最佳化，並且實際應用在影像辨識上。原始作者的參數定義一組固定的參數（如：總共幾張子圖、每張子圖的大小、幾組特徵向量等等），以及使用固定值0.02當作判別兩張圖片是否相似或相異的門檻值。原作者將預先設定的參數全部送入奇異分解演算法內進行運算，最後產生的影像雜湊值差異率如果高於門檻值則判定為兩張不同的影像。經過實驗結果，單單使用固定的參數組，不夠安全也不具有強健性，不足以適用在所有影像上面。

本文前半部使用實驗法則來探討不同的參數定義對奇異值分解演算法所造成的影響，然後根據實驗結果，找尋適合的參數定義。第二、三章我們使用標準影像與自己拍攝的影像當作測試樣本來做出一系列的實驗數據，根據數據顯示經由實驗法則所導出的參數用於奇異值分解演算法所算出來的影像雜湊值會比原始作者固定式的參數定義來的好，並且同時兼具更好的安全性與強健性。

本文後半段使用基因演算法作為參數設定的優化演算法，透過基因演算法的前置處理後，解決了以往使用經驗法則因為隨機臆測的低效率以及欠缺能否達到安全性與強健性優化的把握度問題。第四、五章使用跟前面章節同樣的圖庫進行實驗，並且綜合整篇文章的實驗數據，做出比較表格，證實經由基因演算法確實能使整個過程更有效率，亦能同時兼顧安全性與強健性。

A perceptual image hashing maps an image to a short binary string based on an image's appearance to the human eye. It has to meet two important requirements, namely security and robustness. Security is to detect malicious tampering or modification in images correctly, and robustness is to tolerate incidental modification in images such as rotating, scaling, and transformation.

This thesis makes use of a useful image hashing program tool by Vishal Monga to explore a better parameter set for singular value decomposition （SVD） hashing function used in image authentication. One of the functions provided in Monga’s tool is a SVD based image hashing which currently uses a predefined parameter set to compute the image hashing. However, Monga’s approach starts with an arbitrary threshold constant of 0.02 value to judge the similarity or dissimilarity between two given images, and a predefined parameter set of other parameters （e.g. partition size, sub-image size, and eigenvector number）, which may not be optimal or suitable for generating a secure and robust image hashing for all general images.

First, we cut-and-try a different parameter set to compute the image hash for each image using this SVD algorithm in order to enhance the robustness and security of this algorithm. We present our experiment results in Chapters 2 and 3. It shows the optimal parameter set derived by us has a better performance than Monga’s predefined set. It also constantly shows our approach generates a more secure and robust image authentication when it was tested using the standard test images provided by the USC-SIPI image database.

Next, Genetic Algorithm （GA） given in Chapter 4 is utilized as a preprocessing step to replace the cut-and-try approach in selecting an optimized parameter set for the SVD algorithm. The experiment results listed in Chapter 5 shows our approach works better than Monga and even the redefined parameter sets derived by our cut-and-try method. It clearly demonstrates that SVD optimized by GA technique is an automatic, efficient, secure and robust scheme for image authentication application.

[1]J. J. K. Ò. Ruanaidh and T. Pun, "Rotation, scale and translation invariant spread spectrum digital image watermarking," Signal Processing, vol. 66, pp. 303-317, 1998.
[2]S. Pereira and T. Pun, "Robust template matching for affine resistant image watermarks," IEEE Trans Image Process, vol. 9, pp. 1123-9, 2000.
[3]P. Bas, et al., "Geometrically invariant watermarking using feature points," Image Processing, IEEE Transactions on, vol. 11, pp. 1014-1028, 2002.
[4]K. Mikolajczyk and C. Schmid, "Scale \& Affine Invariant Interest Point Detectors," Int. J. Comput. Vision, vol. 60, pp. 63-86, 2004.
[5]V. Monga and B. L. Evans, "Perceptual image hashing via feature points: performance evaluation and tradeoffs," IEEE Trans Image Process, vol. 15, pp. 3452-65, Nov 2006.
[6]H. S. Prasantha, et al., "Image Compression Using SVD," in Conference on Computational Intelligence and Multimedia Applications, 2007. International Conference on, 2007, pp. 143-145.
[7]F. van Belzen, et al., "Singular value decompositions and low rank approximations of multi-linear functionals," in Decision and Control, 2007 46th IEEE Conference on, 2007, pp. 3751-3756.
[8]J. Holland, Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence: The MIT Press, 1992.
[9]N. Chaiyaratana and A. M. S. Zalzala, "Recent developments in evolutionary and genetic algorithms: theory and applications," in Genetic Algorithms in Engineering Systems: Innovations and Applications, 1997. GALESIA 97. Second International Conference On (Conf. Publ. No. 446), 1997, pp. 270-277.
[10]D. Wicker, et al., "The Multi-Tiered Tournament Selection for evolutionary neural network synthesis," in Combinations of Evolutionary Computation and Neural Networks, 2000 IEEE Symposium on, 2000, pp. 207-215.
[11]Z. Yuanping, et al., "Dynamic Load Balancing Based on Roulette Wheel Selection," in Communications, Circuits and Systems Proceedings, 2006 International Conference on, 2006, pp. 1732-1734.
[12]P. M. Godley, et al., "Fitness directed intervention crossover approaches applied to bio-scheduling problems," in Computational Intelligence in Bioinformatics and Computational Biology, 2008. CIBCB '08. IEEE Symposium on, 2008, pp. 120-127.
[13]S. Sheng and Z. Dechen, "New Genetic Algorithm for the Fixed Charge Transportation Problem," in Intelligent Control and Automation, 2006. WCICA 2006. The Sixth World Congress on, 2006, pp. 7039-7043.
[14]W. Wei-Yen, et al., "GA-based learning of BMF fuzzy-neural network," in Fuzzy Systems, 2002. FUZZ-IEEE'02. Proceedings of the 2002 IEEE International Conference on, 2002, pp. 1234-1239.
[15]L. Guangming and Y. Xin, "Analysing crossover operators by search step size," in Evolutionary Computation, 1997., IEEE International Conference on, 1997, pp. 107-110.
[16]C. P. Chen, et al., "Path optimization using Genetic Algorithm in laser scanning system," in Information Technology, 2008. ITSim 2008. International Symposium on, 2008, pp. 1-5.
[17]C. Pitangui and G. Zaverucha, "Improved natural crossover operators in GBIVIL," in Evolutionary Computation, 2007. CEC 2007. IEEE Congress on, 2007, pp. 2157-2164.
[18]X. B. Hu and E. Di Paolo, "An efficient Genetic Algorithm with uniform crossover for the multi-objective Airport Gate Assignment Problem," in Evolutionary Computation, 2007. CEC 2007. IEEE Congress on, 2007, pp. 55-62.
[19]H. Chao-Jung, et al., "A Modified Genetic Algorithm for Parameter Estimation of Software Reliability Growth Models," in Software Reliability Engineering, 2008. ISSRE 2008. 19th International Symposium on, 2008, pp. 281-282.
[20]N. Gomez Bias, et al., "Networks of evolutionary processors with a self-organizing learning," in Computer Systems and Applications, 2008. AICCSA 2008. IEEE/ACS International Conference on, 2008, pp. 917-918.
[21]S. Abedi and R. Tafazolli, "Genetically modified multiuser detection for code division multiple access systems," Selected Areas in Communications, IEEE Journal on, vol. 20, pp. 463-473, 2002.
[22]W. S. Chen, et al., "A novel constraint non-negative matrix factorization criterion based incremental learning in face recognition," 2008, pp. 292-297.
[23]C. Lazar and A. Doncescu, "Non negative matrix factorization clustering capabilities; application on multivariate image segmentation," 2009, pp. 924-929.
[24]O. Çeliktutan, et al., "Multi-attribute robust facial feature localization," 2008.
[25]T. Ming, et al., "Watermarking technique resisting to strong cropping," 2008, pp. 357-361.
[26]N. K. Batcha and A. M. Zaki, "Preliminary Study of a New Approach to NMF Based Text Summarization Fused with Anaphora Resolution," in Knowledge Discovery and Data Mining, 2010. WKDD '10. Third International Conference on, 2010, pp. 367-370.
[27]N. Bertin, et al., "Fast bayesian nmf algorithms enforcing harmonicity and temporal continuity in polyphonic music transcription," in Applications of Signal Processing to Audio and Acoustics, 2009. WASPAA '09. IEEE Workshop on, 2009, pp. 29-32.
[28]Z. Lei and X. Xuezhi, "Chinese text classification method based on NMF," in Test and Measurement, 2009. ICTM '09. International Conference on, 2009, pp. 240-243.
[29]B. Xiaoming and W. Chengzhang, "Revised NMF with LDA based color face recognition," in Networking and Digital Society (ICNDS), 2010 2nd International Conference on, 2010, pp. 156-159.
[30]Z. Ying and G. Zhang, "Facial Expression Recognition Based on NMF and SVM," in Information Technology and Applications, 2009. IFITA '09. International Forum on, 2009, pp. 612-615.
[31]M. Fatourechi, et al., "Towards automated image hashing based on the Fast Johnson-Lindenstrauss Transform (FJLT)," in Information Forensics and Security, 2009. WIFS 2009. First IEEE International Workshop on, 2009, pp. 121-125.
[32]L. Xudong and Z. J. Wang, "Fast Johnson-Lindenstrauss Transform for robust and secure image hashing," in Multimedia Signal Processing, 2008 IEEE 10th Workshop on, 2008, pp. 725-729.
[33]L. Xudong and Z. J. Wang, "Reduced-reference image quality assessment based on perceptual image hashing," in Image Processing (ICIP), 2009 16th IEEE International Conference on, 2009, pp. 4361-4364.