APPLICATIONS OF DIGITAL SIGNAL-PROCESSING, FUZZY-LOGIC, AND NEURAL NETS TO COMPUTATIONAL COLOR

The matrix-vector approach to computational color allows easy applicat ion of the theory of digital signal processing (DSP). The applications demonstrated to date include color correction, color filter design, a nd improved interpolation and extrapolation methods. Combined with the power of new commercial software, higher computational speed of perso nal computers, and the availability of regional supercomputing centers , practical optimization problems involving nonlinear transformations such as CIELab can be solved in a short time. In addition to more comm on DSP methods, results in fuzzy logic and neural nets which have been applied to signal processing and control problems, can also be easily applied to computational color. Fuzzy set theory allows the natural i nclusion of uncertainty in a color problem, for example the variation of individual observers from the CIE standard observer. Neural nets ca n be used to characterize processes for which there are no adequate ma thematical models. These capabilities are of particular interest in th e determination of the visual effects of the new fluorescent lamps whi ch have recently been introduced.