Multi-modal image segmentation using a modified Hopfield neural network

摘要

In most computer vision applications, it is required to segment objects from a background. In case of a muti-modal image the segmentation is an involved problem in comparison to a bi-modal image. This paper deals with an adaptive technique for choosing local threshold values for faithful image segmentation. The image segmentation is done by generating a threshold surface which is determined by interpolating the image gray levels at points where the gradient is high, indicating probable object edges. The interpolation of edge points is done using a modified Hopfield neural network and the results are compared with that of a potential surface interpolation method.