Texture image retrieval based on complex wavelet and relative entropy
Fangnian Lang · Journal of Chongqing University. English Edition · 2008
In order to enhance the performance of the texture image retrieval,a new method based on the complex wavelet transform(CWT) was presented.Using the wavelet coefficient histogram distribution of the CWT real and imaginary part as the texture feature,the method was developed by combining the characteristics of the complex wavelet decomposition and feature extraction with similarity measurement of relative entropy for image retrieval.In comparison with a traditional wavelet,this method is better than the pyramid discrete wavelet decomposition transforms(PDWT) with the same feature extraction method and similarity measure.In the contrast experiment results for image retrieval,the retrieval efficiency of the CWT is better than that of the PDWT.The efficiency of Kingsbury's method is superior than that of Fernandes's within the CWT in the texture retrieval.