A neural network for enhancing boundaries and surfaces in synthetic aperture radar images

A neural network system for boundary segmentation and surface representatio n, inspired by a new local-circuit model of visual processing in the cerebr al cortex, is used to enhance images of range data gathered by a synthetic aperture radar (SAR) sensor. Boundary segmentation is accomplished by an im proved Boundary Contour System (BCS) model which completes coherent boundar ies that retain their sensitivity to image contrasts and locations. A Featu re Contour System (FCS) model compensates for local contrast variations and uses the compensated signals to diffusively fill-in surface regions within the BCS boundaries. Image noise pixels that are not supported by BCS bound aries are hereby eliminated. More generally, BCS/FCS processing normalizes input dynamic range, reduces noise, and enhances contrasts between surface regions. BCS/FCS processing hereby makes structures such as motor vehicles, roads, and buildings more salient to human observers than in original imag ery. The new BCS model improves image enhancement with significant reductio ns in processing time and complexity over previous BCS applications. The ne w system also outperforms several established techniques for image enhancem ent. (C) 1999 Elsevier Science Ltd. All rights reserved.