TEXTURE CLASSIFICATION USING WEIGHTED PROBABILISTIC NEURAL NETWORKS

Texture classification is basically the problem of classifying pixels in an image according to their textural cues. This is different from conventional image segmentation as the texture is characterized using both the gray value for a given pixel and gray-level pattern in the neighborhood surrounding the pixel. In this project, the novel temporal updating approach is developed for weighted probabilistic neural network (WPNN) classifiers that can be used to classify the textures. This is done by utilizing the temporal contextual information and adjusting the WPNN to adapt to such changes. Whenever a new set of images arrives, an initial classification is first performed using the WPNN updated to the last frame while at the same time, a prediction using PNN is also based on the classification results of previous frame. The result of both the PNN and WPNN are then compared. Compared to the PNN, WPNN includes weighting factors between pattern layers and summation layer of the PNN. Performance of this approach is compared with model based and feature based methods in terms of signal to noise ratio and classification rate