Image Classification System Based on Multi-wavelet and Neural Network

Abstract

Image classification with the generation of signature to the image. This paper propose approach of image classification based on multi-wavelet transform which has a matrix structure, and combines plays a great role in industrial, remote sensing, and military applications. It is concerned all features that a simple scalar wavelet cannot have at once. A successful classification rate of 98% was achieved with this method.Also, the Neural Network (NN) classifier is combined with the multi-wavelet transform. The purposed neural network is a feed-forward multilayer. The learning algorithm uses is Levenberg Marquardt Algorithm. It is found that such combination is capable of performing advanced computation and approximates of a desired input-output behavior. The effect of this combination on the classification task is considered an average classification rate of 100% is achieved for Discrete Multi-wavelet with Neural Network method.