The theory and experiment of a new white-light image-processing technique for black-and-white image addition and subtraction by means of Fourier-plane spatial color filtering are described. Both the intensity transmittances of three (or two) processed images and the results of the addition and subtraction are expressed in a chromaticity coordinate system.

A new technique of one-step rainbow holography with a multiexposure holographic lens (MEHL) is described. The synthesization of the slitlike function in the object wave can be performed with the specially fabricated holographic lens. The MEHL can also be applied to the oriented speckle encoding of images in white-light optical image processing. Successful experimental results are presented.

An algorithm with non-linear weight factors in the summation process for fuzzy correlation of color his-tograms is presented, in which non-linear weights are assigned to some characteristic colors of interest. Experimental results show that this can improve the retrieval of color images with partial aberrations or with local color characters.

Theoretical and experimental analyses of color-image pattern recognition using a multiwavelength Fresnel holographic filter (FHF) are presented. The FHF technique permits a greatly relaxed spatial-filter alignment requirement, use of a white-light source, negligible color cross talk, and high diffraction efficiency of the spatial filter.

A method of amplitude-compensated matched filtering is presented. This yields much better discrimination and a sharper autocorrelation peak compared with the classical matched spatial filter and the phase-only one. By computer simulation in the case of alphanumeric characters, it is concluded that, although the phase information is significantly more important, in the case of a matched filter, the amplitude is important too. Three-dimensional graphs of autocorrelation and cross-correlation are also provided.