Selection method of optimum wavelet bases for image signature
Xiaofeng Wang, Chunni Dou, Lihua Tian · 2010
The properties of wavelet bases have greatly influence on algorithm performances when wavelet functions are used to construct image signature. Characteristics of image signature, such as robustness, tamper sensitivity, tamper localization, and compactness are analyzed in this work. Moreover, properties of Daubechies wavelet and its decomposition levels are analyzed. According to the results of our study, image signature generated by dbl, db2, or db3 show stronger robustness and compactness yet lower tamper sensitivity and tamper localization capability with a higher wavelet decomposition level; and it exhibit higher computational complexity and lower compactness with a higher wavelet vanishing moment. Our results suggest that db2 wavelet is suitable for the construction of image signature, and also, its decomposition level has been recommended.