EQUAL CORRELATION PEAK OPTIMIZATION OF THE FILTER-FEATURE-BASED SYNTHETIC DISCRIMINANT REFERENCE IMAGE

A modified equal correlation peak (ECP) optimization is proposed and i ntroduced into the filter-feature-based joint transform correlator (FF B JTC) used for distortion-invariant pattern recognition. Instead of o ne fixed and arbitrarily chosen point in the classical ECP optimizatio n, the maximum point in the correlation output plane is selected in ou r method. The modified ECP optimization is quite efficient when involv ing filter modulation. When compared with other ECP optimization metho ds related to filter-modulation, our optimization method is linear and can easily obtain an exact solution. Simulation results show that aft er ECP optimization the stability of the correlation output of the FFB JTC may be greatly improved. (C) 1998 Elsevier Science B.V.