An effective way of image compression using DWT and SOM based vector quantisation

Vector quantisation is a novel and attractive technique for still image compression. The most significant advantages are high reconstruction quality at low coding rates, rapid decoding. One of the major disadvantages is high encoding time complexity. Different algorithms are being developed to reduce the search space, so that encoding time complexity can be reduced to a large extent. An algorithm based on Kohonen´s self organizing maps is proposed in this paper for quantising the image data. This algorithm uses its roots in neural networks based on neighbourhood relationships. Wavelet transform is also a cutting edge technology in the field of image compression. This provides substantial improvement in picture quality at high compression ratios. Experimental results obtained demonstrate the effectiveness of the proposed algorithm.