A fringe-adjusted joint transform correlator (JTC) based technique for improved color pattern recognition is introduced. in the proposed technique, a real-valued fringe-adjusted filter is used to reshape the joint power spectrum to yield better correlation output. A color image is processed through three channels, and fringe-adjusted filtering is applied to each of these channels to achieve excellent correlation discrimination. The correlation outputs from these channels are then fused together to make a decision on the detection of a desired color pattern. It is also shown that the fringe-adjusted filtering can be effectively applied to a multichannel single-output JTC to obtain excellent correlation discrimination between an unknown input scene target and a reference image for all color channels. The proposed techniques can be implemented in real time for practical color pattern recognition applications. Two architectures for all-optical implementation of the proposed techniques are presented.