Vector quantization of color digital images within a human visual model

A method of vector quantization which uses a color model of the human visual system to improve the quality of encoded images is presented. In this method, the images are first transformed into a 'perceptual space' and then encoded. After reconstruction, the image is transformed back into an intensity representation for viewing. A comparison is made between images encoded at 1.125 bits per pixel with and without the visual model using both an SNR measure and visual observation. Experimental results indicate that there is a significant improvement in the perceived quality of the images encoded in the visual model, although the measured distortion between the images is slightly better for the intensity encoded images.<
>