Abstract
In this paper, geometric invariant texture analysis, which comprises of rotation, scale and translation (RST) invariance, is presented. Many types of moments and functions of moments have been utilized in RST invariant pattern recognition applications. However, use of moments for texture content-based image analysis is limited. Here, application of Zernike moment for geometric invariant texture analysis is studied. An algorithm is proposed to design fast and modified Zernike moments to extract a particular texture from image irrespective of its rotation scale change and translation. The algorithm is tested for its robustness in the presences of various noises like salt and paper, Gaussian and speckle noise. The operator is evaluated for computation of moments on different data sets of textures like Brodatz and Vistex. Comparison is done with traditional Zernike moments based techniques, Fast Fourier Transform (FFT)-Zernike moments based technique and proposed modified Zernike moment based technique.