A modified gray-level difference algorithm for analysing Gaussian Blurred texture images

A modified texture feature extraction method based on gray-level difference (GLD) algorithm is presented in this paper. Briefly, the proposed method is composed of four steps: 1) the variance of the image´s texture part is estimated and compared to the preset threshold; 2) the Wiener filter was applied to remove Gaussian blur noise iteratively if the variance is lower than the threshold; 3) until the variance is higher than the threshold and then iteration stops; 4) the conventional GLD algorithm is used to extract the texture information from the processed texture image. To measure the discrimination performance of the new algorithm, Experiments are operated and the results indicate that the new method is better than the conventional GLD algorithm while extracting textural information from image with noise, especially when contaminated by Gaussian blur noise.