Recognition of Degraded Traffic Sign Symbols Using PNN and Combined Blur and Affine Invariants

A fast version of probabilistic neural network model is proposed. The model incorporates the J-means algorithm to select the pattern layer centers and genetic algorithm to optimize the spread parameters of the probabilistic neural network, enhancing its performance. The proposed approach is applied to the recognition of degraded traffic signs with promising results. In order to cope with the degradations, the Combined Blur and Affine Invariants (CBAIs) are adopted to extract the features of traffic sign symbols without any restorations which usually need a great amount of computations. The simulation results indicate that the fast version of PNN optimized with GA is not only parsimonious but also has better generalization performance.