The Use of Run Length and Contrast Features with Neural Network for Texture Recognition

The analysis of texture parameters is a useful way of increasing the information obtainable from images. It is an ongoing field of research, whether these images are medical images or natural, with applications to characterizes the variation in textures and classify the texture. In this research, a comparative study of conventional texture-analysis methods and the proposed methods. The textural features extracted by these methods are exploited to classify regions of interest (ROIs). Four different sets of features are proposed the first set is a simple modified features extracted from the traditional Run Length Matrix (GLCM), the second set is features extracted from Contrast Matrices (CM), the third set uses features extracted from combination between Contrast Matrices (CM) and run length feature vector, the fourth way was passing the extracted sets of these previous proposed set through Artificial Neural Network (ANN) for classification purpose. It proved that the proposed methods are superior to the conventional texture-analysis method with respect to classification accuracy and computational complexity.