抄録
A T R Auditory and Visual Perception Research Laboratories ABSTRACT Image data compression techniques are fundamentally based on the suppression of redundancy, maintaining an acceptable image quality. In this paper, the applicability of Neural Networks to image coding is discussed. We proposed a local-parallel type multilayer Neural Network model which can get abilities of image data compression and reconstruction by Back-propagation learning rule. Computer simulation experimentally shows that image compression and reconstruction ability can be realized on Neural Network for natual facial images.
