Areas of Application
(See summary below; then, click on blue panels for details)
The photoreceptor layer in the human retina is, in very general terms, like a CCD array of light-sensing elements. The powerful day-time human vision starts with signals from a single layer of 3 types of color sensitive photoreceptors. Therefore, at each retinal "pixel" there is initially only one color primary signal, R, G, or B. These signals interact in the parvocellular neural networks in the retina and at higher levels of the brain to endow vision with both its finest spatial resolution and color, with precision down to the single retinal "pixel" level. Chromaplex's algorithms are based in those neural interactions. They are designed to perform calculations similar to the center-surround color opponent neurons subserving the fovea, which can be conceived as computing the other two color primaries missing at each retinal "pixel".
The engineering application of visual multiplexing embodied in Chromaplex was developed under the realization that a single array of signal values of 3 types (RGB) can be processed, compressed, stored, and transmitted with the simplicity of a monochrome image, after we solved the problem, for high compression rates, of the additional entropy introduced by having three types of values in the same array. Then, when the image is to be displayed, very simple neuron-like calculations (see the fourth reference below) provide the two missing color values at each point in the array, producing a full color image of high quality and fine spatial resolution.
Other technologies do not yield images of similar quality because they are based on interpolation operations between values of the same type (either R, G or B), or based on cross-correlation operations between colors that fail to incorporate the advantages of the biological processes of visual neurons discovered at SRI. This is evident in the generalized use of blur filters (optical or computational) in digital color cameras based on a single CCD chip, and the addition of algorithms for image restoration which are computational expensive and solve only a limited amount of artifacts.
For information on specific areas of application and image demonstrations, click on the blue panels at the top of this page.
Martinez-Uriegas, E., H. D. Crane, and J. D. Peters. "Method and apparatus for compression and decompression of digital color images". United States Patent No. 5,398,066. March-14-1995. Abstract
Peters, J. D., H. D. Crane, and E. Martinez-Uriegas. "Method and apparatus for increasing resolution of digital color images using correlated decoding". United States Patent No. 5,541,653. July-30-1996. Abstract
Martinez-Uriegas, E. "Chromatic-achromatic multiplexing in color vision". Chapter 4 In: Kelly, D.H. (Ed.) "Visual Science and Engineering", 117-187. Marcel Dekker, New York. 1994.
Martinez-Uriegas, E., H. D. Crane, and J. D. Peters. Spatiochromatic multiplexing: A color image representation for digital processing and compression. SPIE Proceeding Series, Vol 2657, 412 - 420, 1996.
Chromaplex is a trademark of SRI International
