Real time detection of traffic signs
El Hafidi, Badr
MetadataShow full item record
The objective of this research project is to develop a high-performance real-time algorithm for traffic sign detection using a colorful image. The detection of traffic signs is achieved through two steps: the first step extracts the regions of the same color as the traffic signs using an improved color segmentation filter and the second step identifies the shape of the extracted region using a shape recognition algorithm. Classic color segmentation approaches use only one color-range. While they work properly with good weather conditions and high-quality images, they usually struggle with illumination changes, shadow, highlight and harsh weather conditions. Moreover, they can’t extract the traffic sign from the background if the latest is in the same color range of the traffic sign. The proposed work outlines an approach that combines both shape and color and uses an improved color segmentation algorithm. It uses a specific color space, which is HSV space, and then builds a set of HSV domains that each corresponds to a specific lightening condition. The approach developed in this work demonstrated the invariance of the detection to shadow, light conditions and weather conditions.