We propose three different methods for pattern recognition that use a joint transform correlator architecture. They are based on the use of a liquid crystal television operating in a phase-mostly mode along all the process. This feature makes these methods suitable to work with low levels of illumination. In the first step of the joint transform correlation the scene and the reference are phase encoded. In the second step we obtain the correlation function by three different ways. The first one is based on the phase encoding of the joint transform spectrum. In the second one the interference term of the joint spectrum is binarized, and in the third one we extract the phase distribution of the joint transform spectrum. The last method allows us to obtain the pure phase correlation. These methods are experimentally implemented and the results are compared with those obtained with the amplitude binarization method. Effects of quantization and saturation are numerically simulated for the phase extraction method. © 1998 Elsevier Science B.V. All rights reserved.
|Publication status||Published - 15 May 1998|
- Joint transform correlator
- Liquid crystal TV
- Pattern recognition