The focus of this paper is the emulation of the visual receptive fields from the retina to the visual cortex to extract information form images for perceptual sub-band coding. The paper introduces a generalized family of differentio-aggregation (DA) functions which emulate the biological visual receptive fields for information extraction from 2D spatial images. These functions are able to extract the transition information in the presence of noise very distinctly. Presently there are a limited set of image processing masks available for this purpose and moreover successive processing of images with these masks does not make sense. A method to use these functions to perform perceptual subband coding is outlined. Several different types of DA functions are employed to extract transition information from a medical image. Results from processing a spine X-ray are presented, and different sub-band coded images are included to illustrate the usefulness of the DA functions for coding applications.
展开▼
