Method of Traffic Signs Segmentation Based on Color-Standardization

The drawbacks of commonly applied color treatment based on the HSI space model have been identified as being the additional computation load for color model conversion, which results in compromising the real-time traffic sign detection. The relevant analysis shows that the major factor adversely affecting the segmentation of traffic signs is closely associated with the color complexity. To reduce the computational complexity in the scene image processing, a new solution based on the color standardization, as defined in this work, is proposed. The scene image is mapped to a simple standardized image consisting of standard color of eight categories, among which five colors related to traffic signs are extracted to form the standardized traffic signs region. The experimental results show that the proposed approach simplifies the complex color description of the scene image, and achieves the traffic signs segmentation with sufficiently high processing speed and satisfactory accuracy.