A Low Indexed Content Based Neural Network Approach for Natural Objects Recognition

In this paper, an approach to integral color texture invariant information with a neural network approach to object recognition is proposed. A color-texture context for image retrieval system based on the integral information of an image is represented as one compact representation base on color histogram approach. A general and efficient design approach using a neural classifier to cope with small training sets of high dimension, which is a problem frequency encountered in object recognition, is focused in this paper for general images. The proposed system is tested for various colored image samples and the recognition accuracy is evaluated.