A neural network based automatic road signs recognizer

Automatic road sign recognition systems are aimed at detection and recognition of one or more road signs from real-world color images. In this research, road signs are detected and extracted from real world scenes on the basis of their color and shape features. A dynamic region growing technique is adopted to enhance color segmentation results obtained in the HSV color space. The technique is based on a dynamic threshold that reduces the effect of hue instability in real scenes due to external brightness variation. Classification is then performed on extracted candidate regions using multilayer perceptron neural networks. The obtained results show good detection and recognition rates of the entire system with real outdoor scenes, using several light conditions. Finally, the implementation of the neural layer on the Georgia Institute of Technology SIMD Pixel Processor is outlined.