HISTOLOGICAL IMAGE DETECTION USING STATISTICAL TESTS AND A NEURAL NETWORK

摘要

Many methods have been developed for image edge detection and most of these techniques work well in images with uniform regions, but less well in regions with greater non-uniformity. This paper describes an edge enhancing technique suitable for the complexity of histological images, such as those contained in the Mouse Embryo Atlas (MA), In order to achieve this, a new feature exaction method has been developed which combines a neural network technique with a novel edge detection algorithm. The neural network has been trained on several sets of appropriate images that classify a selection of pixels as edge or non-edge points. Our preliminary results are promising showing that many edges have been successfully enhanced with few false positives.