A method for traffic signs detection, tracking and recognition

This paper describes a method for traffic signs detection, tracking and recognition. Color and shape are combined to detect signs. Hue and saturation are used to detect the red color of the sign. Circles are detected through the improved round-degree method of extracting area feature parameters. An improved Kalman filter is introduced to track multiple targets in the next frames. A feature extraction method for traffic sign recognition based on Two-Dimensional Principal Component Analysis (2DPCA) is proposed. A series of experiments are performed on two traffic sign databases with the nearest neighbor classifier and Euler distance. One database is the image library that images are obtained through a series of simulation transformation after image binarization. Another database is made up of images shot from real scenes through selecting many different location scenes. The method has a good effect on the recognition of the both image databases.