Illumination-invariant pattern recognition with joint-transform-correlator-based morphological correlation.

The performance of nonlinear morphological correlation is investigated and compared with that of conventional linear correlation. In particular, the effects of illumination variations on the morphological correlation output are investigated in detail. The morphological correlation is shown to be invariant to uniform input-image illumination when the input-image illumination is higher than that of the reference. It also provides higher pattern discriminability, sharper peaks, and more-robust detection in the presence of salt-and-pepper noise than does the linear correlation. Computer-simulation results are provided.