COLOR IMAGE PROCESSING WITH COMPUTER-GENERATED SPATIAL FILTERS IN A DISPERSED WHITE-LIGHT OPTICAL SYSTEM (SIGNAL, INFORMATION, HOLOGRAPHY)

Computer-generated (CG) spatial filters are applied to color image processing in a dispersed white light optical system. The range of input wavelengths produces continuously dispersed Fourier spectra in the filter plane, each having scale factor proportional to wavelength. Analysis of the requirements for multi-, discrete-band filters to achieve real bipolar and complex filtering operations on the dispersed color spectra is presented. Computer simulations, for the linear image deblurring example, show that significant processing performance can be achieved by filters realized with typical CG filter-making devices. Processing performance is more sensitive to filter geometry and alignment for the color processor than for a monochromatic processor, primarily because of color smear effects, similar to those produced with holograms under white light illumination. Experimental results with binary and continuous-tone smeared color images agree well with computer simulation results. Wavelength discrimination in the color image display material largely determines the effectiveness of wavelength as an additional degree of freedom in an optical processing system. Finally, a parallel processing technique is proposed which can increase the number of CG filter processing channels for moderate resolution color input images.