Embedded Zero Tree Wavelet based Artificial Neural Network Image Classification Algorithm - A Study

In this work, an urban area land cover is proposed to classify the large resolution image. It aims to extract the features like texture, shape, size and spectral information in the feature extraction process. Embedded Zero tree Wavelet transform is a lossy image compression algorithm. Most of the coefficients at low bit rates bent through a sub band transform will be zero, or very close to zero. These features data are used for the classification process. Here, we used various classification algorithms namely, Radial Basis Function, SMO, Multilayer Perceptron and Random Forest are implemented. The classification accuracy constantly depends on the efficiency of the extracted features and classification algorithms. The result of the proposed classification algorithms are merged with EZW. Experimental results illustrate that the better accuracy performance is obtained by the Multilayer Perceptron algorithm than other classification algorithms.