Neuro-Statistical Vector Quantizer(NSVQ)Design in The Spatial Domain

Image compression is the process of reducing the number of bits required to represent an image. To enhance the performance of image compression, an efficient and intuitive technique is proposed based on a Neuro- Statistical model taking into account the psycho visual features in the spatial domain. First, Competitive Learning Algorithm is used to generate the codebook. Then, each code vector of the codebook is modelled by the Savitzky- Golay polynomial, for obtaining better perceptual fidelity of the reconstructed images. Simulation results show that the proposed Neuro-Statistical Vector Quantization technique provides better perceptual quality in terms of BSMI and PSNR ratio while maintaining compression ratio similar to that of the existing methods.