A neural network shape recognition system based on D-S Theory
Liangmei Hu, Jun Gao, Andong Wang, H.S. Young · 2004
In this paper, a new neural network shape recognition system based on Dempster-Shafer theory is presented. It is composed of three parts; they are preprocessing part, feature extracting part and recognition part. Firstly, we use Hough Transform (HT) to preprocess and obtain the feature vectors of the images to be recognized. Recognition part fully utilizes the advantages of Dempster-Shafer Theory in uncertainty reasoning, and the prototype patterns are used as items of evidence in Dempster-Shafer reasoning. The belief degrees deduced by those evidences are represented by basic belief assignments (BBAs) and pooled using the Dempster's rule of combination. This procedure can be implemented in a multilayer neural network with specific architecture consisting of one input layer, two hidden layers and one output layer. Experiments in recognition of three kinds of traffic signs demonstrate the excellent performance of this recognition system.