Image Classification based on Color and Texture features using FRBFN network with Artificial Bee Colony Optimization Algorithm

Abstract

With advances in information technology, there is an explosive growth of image databases which demands effective and efficient tools that allow users to search through this large collection. Conventionally, the way of searching the collections of digital image database is by matching keywords with image caption, descriptions and labels. Keyword based searching method provides very high computational complexity and the user has to remember the exact keywords used in the image database. Even though the computational complexities of the traditional image retrieval systems are high, they produce low classification accuracy. This paper proposes an image classification system based on combined color and texture features of an image to overcome these problems. This system consists of different stages such as image preprocessing, color and texture features extraction and fuzzy c-means radial basis function neural (FRBFN) network based classification/retrieval with Artificial Bee Colony (ABC) optimization algorithm. In this scheme, the color features are derived using Histogram Equalization method in HSV space and the texture features represented by contrast, energy, entropy, correlation and local stationary over the region in an image derived based on co-occurrence matrix. The proposed neural network based Comprehensive Image Classification (CIC) scheme fuses the low level features of the image such as color and texture to improve the systems classification performance and these features are converted as high level features by Radial Basis Function Neural Network(RBFN) with Fuzzy c-means (FCM) to fix the hidden layer neurons. The weight vectors of the network are reasonably assigned by Artificial Bee Colony (ABC) optimization algorithm. The experimental results show that the proposed method is superior to other traditional Content Based Image Retrieval (CBIR) methods in classifying the images with less computational time. The performance of proposed scheme was evaluated with a set of 600 color images taken from COREL benchmark image library and the routines of the proposed system were simulated using MATLAB R2008b. General Terms Neural Networks, Image Classification