A. Vavilin, K. Deb, and K.-H. Jo (Korea)
Guidance sign recognition, genetic algorithm and structure analysis.
In this paper we present an approach to detect arrowheads inside of the traffic guidance signs. This method is based on deformation of arrowhead model guided by a genetic algorithm. Initial model consists of two parts: edge mask and shape mask. Each of the masks is generated from multiple input images using neural network. This matter is used to improve robustness of the algorithm for different types of arrowheads. During the search process the model is deformed using five parameters such as scaling, shifting and rotating which form the transformation vector. Each of the vectors represents an individual in population of genetic algorithm. Evolutionary improvement of the population allows precise detection of arrowhead regions inside of the sign. The proposed method is compared with border tracing algorithm and iterative improvement method.
Important Links:
Go Back
