Image compression using enhanced vector quantizer designed with selective training of unsupervised neural network

This paper presents novel approach for compressing color images using vector quantizer designed with self organizing feature maps unsupervised neural networks. The design incorporates selective training of network for reducing blocking artifacts ,which deteriorates image quality. The technique also achieves better trade off between quality and compression. The quality analysis is also done by applying various quality measures. Finally, the comparison with popular JPEG with VQ is presented.