Extracting Circular Road Signs Using Specific Color Distinction and Region Limitation

摘要

There is now vigorous research on the intelligent transport system (ITS), and also on the advanced cruise-assist highway system (AHS), which is one of the ITS components. An element of the AHS is the system that extracts and recognizes road signs from color road images to alert the driver. This paper proposes a new technique that can extract circular road signs in a practical processing time. The proposed method first discriminates the color used in the road sign (specific color) from the input color road image. Then, the target region for extraction is limited on the basis of the color distribution, so that the processing time is drastically reduced. Furthermore, by using the area ratio of the specific color in the extraction region, the extraction accuracy is improved and the processing time is reduced. In evaluation experiment using 53 color road images, a high detection rate of 94.7% was achieved. The average processing time was approximately 2.18 seconds, which is close to practical.