ABSTRACT
Models of lateral inhibition in biological vision systems have long been recognized as sharing both form and function with operations of digital image processing. This chapter describes an implementation of the stripped down Pinter model, discusses its behavior as an image processor, and compares it to more complete implementations. The adaptation to local luminance described by Pinter in his multiplicative lateral inhibition model shows an increased edge enhancement as local mean intensity increases, as well as an overall gain adaptation. The chapter utilizes solutions of the 2D model to illustrate the potential for the Pinter model to be used for facets of image processing. Preliminary experiments of a two dimensional implementation yielded results in qualitative agreement with the one dimensional results and suggested the model's image enhancement potential. Image enhancement uses linear high frequency emphasis spatial filters whose impulse responses are identical in form to the lateral inhibition response profile and whose frequency responses are narrow band spatial frequency channels.
