Experiments in image enhancement using biological and artificial neural networks

In this thesis, we report the results of computer experiments carried out to enhance digital images. We consider two areas of image processing: edge/boundary detection and image restoration. In our research on edge detection, we use a special line-weight function (LWF) which is a combination of zero- and second-order Hermite functions. We are motivated by physiological evidence reported in (78) that visual receptive fields are shaped like the sum of a Gaussian function and its Laplacian. This function can also be derived mathematically when the contrast sensitivity experiments of psychophysics are posed as a eigenvalue problem (71). We provide experimental and mathematical proof that the LWF produces only authentic scale space contours (i.e. it does not detect phantom edges). We also show that the LWF has an excellent localization property. Performance comparison between the Laplacian of Gaussian and LWF operators is presented in a number of computer simulations.; We extend our work from edge to boundary detection through the use of projection pursuit learning networks (PPLN). PPLNs have been used in many fields of research but have not been widely used in image processing. We demonstrate how a PPLN may be used to produce more continuous boundaries when presented with broken edge segments. We also propose the application of PPLN to deblurring a degraded image when little or no a priori information about the blur is available. The PPLN was successful at developing an inverse blur filter to enhance blurry images.