Visual Communication - Visual Communication with Retinex Coding
Visual Communication with Retinex Coding
The inclusion of retinex processing in an informationally optimized image system involves two steps:
The retinex transformation uses an algorithm for enhancing color images that Jobson et al.1,2 developed from a model of color constancy in human vision. To characterize and optimize retinex imaging in terms of information theory, it is necessary to develop an appropriate small-signal representation of the nonlinear retinex-encoded signal. This representation shows that retinex imaging consists essentially of the familiar difference-of-Gaussian bandpass filter and a locally adaptive automatic-gain control. The suppression of shadows in the restored image is found to be constrained inherently more by the sharpness of their penumbra than by their depth.3 Current activities are directed toward color and multispectral retinex imaging and to more general assessments of nonlinear coding in terms of information theory.
- Signal coding with the nonlinear retinex transformation to
- enhance edges sharply and clearly, like the Mach bands in human vision, free from the compromise between sharpness and ringing due to bandlimited linear processing
- preserve the contrast across these edges relative to the mean reflectance of their neighborhood.
- Image restoration with a linear Wiener filter to minimize the mean-square error due to the perturbations by
- blurring, aliasing and photodetector noise in image gathering
- errors in suppressing the irradiance and preserving the reflectance.
- D. J. Jobson, Z. Rahman and G. A. Woodell, "Properties and performance of a center/surround retinex," IEEE Trans. on Image Proc. 6, 451-462 (1997).
- D. J. Jobson, Z. Rahman and G. A. Woodell, "A multi-scale retinex for bridging the gap between color images and the human observation of scenes," IEEE Trans. on Image Proc. 6, 965-976 (1997).
- F. O. Huck, C. L. Fales, R. E. Davis and R. Alter-Gartenberg, "Visual communication with retinex coding," Appl. Opt. 39, 1711-1730 (2000).
Web site curator: Glenn Woodell