WAVELET-BASED IMAGE COMPRESSION USING SUPPORT VECTOR MACHINE LEARNING AND ENCODING TECHNIQUES

摘要

This paper presents a method of compressing still images combining the powerful features of support vector machine (SVM) for machine learning with discrete wavelet transform (DWT) in image transformation. DWT, based on the 'haar' wavelet, has been used to transform the image and the coefficients acquired from DWT are then trained with SVM using Gaussian kernel. SVM has the property that it selects a minimal number of coefficients to model the training data for a predefined level of accuracy. The coefficients are then quantized and encoded using the Huffman coding algorithm. The performance of the proposed method is aspiring and comparable with the existing image compression standards.