Color Image Filtering Using Convolution Fuzzy Neural Network

In this chapter, color image filtering using convolution fuzzy neural network is dealt with. Different types of filters are applied, and the resultant images are obtained using MATLAB. The original image is fuzzified, and the corresponding membership image is found. The image obtained by applying Gaussian filter for different membership and crisp values is depicted. The procedure of convolution operation using Gaussian filter is explained. Features of the images are taken as trapezoidal fuzzy numbers. Hue- and saturation-modified images for different membership values are found. The images for different types of filters such as mean, median, bilateral, Lab, noise, unsharp, standard, Gaussian filters, and the corresponding membership images are obtained. Hamming and Euclidean distances between filtered images and corresponding membership images are calculated, and using these distances, similarity index is evaluated. It is found that images acquired by using mean and bilateral filters are almost similar.