Optical image processing: The current state of the art

Optics have many attractive properties, such as high speed, massive parallelism and non-interference. Optical (especially analog) technology has been used for years in applicadons such as: image processing using spatial filtering of coherent light, pattem recognition using matched filters, signal processing using acoustooptic devices and matrix-vector multiplication using discrete optical processors. However, in recent years, with the maturing of laser technology, research results have indicated that digital optical processing is feasible and may offer considerable benefits in terms of speed and parallelism. In this paper we present an assessment of optical computing as applied to the domain of image processing and pattem recognition, and how newly developed techniques such as symbolic substitution and sptial-temporal integration can be used much effectively than the conventional image processing techniques for electronic computers.