Image Recognition of Traffic Speed Signs in Scene Image Using Genetic Algorithms and Neural Networks.

Abstract

We present a new technology for recognition of traffic signs in a scene image using genetic algorithms and neural networks. Although human beings have an excellent faculty of pattern recognition, the process of pattern recognition has not yet been clarified. Numerous studies have been conducted to realize computer vision similar to that of humans using image processing technology. However, if factors such as position, size, and background of objects are not distinct in the image, the recognition is difficult. In this study, by application of genetic algorithms, a new method is proposed for recognition of objects from a scene image using only the brightness. First, the original image is converted to binary image using a smoothing filter and a laplacian filter. Then, we locate the traffic sign using the proposed genetic algorithm by analyzing both position and size information. Next, the second traffic sign is detected by convergence condition of individual. Finally, we use neural networks to identify the detected traffic sign. These experimental results shows that the new technology proposed here is capable of recognition of traffic signs from a scene image.