Abstract

Image processing often employs sophisticated digital algorithms or precisely aligned coherent optical techniques. Although these methods are typically efficient in their image processing capabilities, neither method readily lends itself to simple, real-time implementation. For binary image processing, the importance of optical morphological filtering has recently been recognized.^1,2 Through the optical implementation of mathematical morphology, systems capable of performing operations, such as median filtering, skeletonization, edge extraction, and impulsive noise removal have been studied. Though often real-time in nature, these systems usually require complex optical arrangements or nonlinear electronic-feedback networks. We present an edge-extraction technique for binary images that implements morphological transformations through the use of a technically simple hybrid optical/electronic system. The optical portion of our system consists of an optical-scanning arrangement that requires only minimal optical alignment. The electronic portion of our system uses simple analog circuitry, which requires no feedback and eliminates the need for sophisticated digital/electronic processing. The same system is also used to extract edges from gray-scale objects by subtracting binary slices obtained through a threshold-decomposition operation.² The processing speed of our system is essentially limited by how fast we can scan an object. We have obtained good results in extracting edges in real time from both binary and gray-scale objects.

© 1990 Optical Society of America