Design of Fusion Texture Feature with Orthogonal Polynomials Model and Co- Occurrence Property for Content Based Image Retrieval

In this paper, a new fusion texture feature with orthogonal polynomials based multiresolution subband and the Gray Level Co-occurrence Matrix (GLCM) is presented. The proposed orthogonal polynomials based multiresolution subband coefficients posses the localized frequency information and the GLCM matrices capture the structural and statistical properties from the subband coefficients for characterizing the texture features. A set of texture features is derived and is experimented with the popular texture database Broadtz for texture image retrieval. The experiment shows that the proposed fusion texture feature outperforms well for the regular, irregular and weak directionality texture images. The proposed method yields high retrieval rate when compared with Discrete Wavelet Transform (DWT) based retrieval scheme with less computational cost.