The present paper is an extension of our previous paper [1]. In this paper shape descriptors are derived on binary cross diagonal texture matrix (BCDTM) after formation of morphological gradient on the wavelet domain. Morphological gradient is obtained from the difference of dilated and eroded gray level texture. A close relationship can be obtained with contour and texture pattern by evaluating morphological edge information. Morphological operations are simple and they provide topology of the texture, that is the reason the proposed morphological gradient provides abundance of texture and shape information. The proposed Wavelet based morphological gradient binary cross diagonal shape descriptors texture matrix (WMG-BCDSDTM) using wavelets is experimented on wide range of textures for classification purpose. The experimental results indicate a high classification rate.