Abstract/Summary

Modified probabilistic neural networks are introduced for image classification based on various distance measures in probability space, in which the input to the model is local binary pattern histogram of images. Conventional probabilistic neural networks have input layer, which computes Euclidean distance of pairwise input features. The proposed modified probabilistic neural networks considered various probability distance measures for computing distances of LBP histograms between images. Extensive comparative experimental studies are employed to demonstrate the effectiveness of the proposed method. It is shown that the probabilistic neural network (PNN) based on Bhattacharyya distance measure is superior to other measures and using the subset of uniform LBP features is generally better than using full LBP features.