A Model of Human Feature Detection Based on Matched Filters

It is generally accepted that edge and line detection is an important stage of any visual system, biological or artificial. Many algorithms have been developed, either to simulate how humans may detect lines and edges, or as a stage in artificial image processing (see Hildreth, 1985) . Most algorithms convolve the input image with operators of limited bandwidth, and search either for zero-crossings or peaks in the output.