An Automatic Indian Traffic Signs Detection and Recognition System Using OpenCV

The main objective of this research is to develop an automatic Indian traffic signs detection and recognition (TSDR) system which is fast and efficient for real-time implementation in autonomous vehicles. A TSDR system is an essential component of autonomous vehicles and a fast algorithm is necessary for its implementation in a real-time environment. This paper presents a TSDR system developed using Open Source Computer Vision Library (OpenCV) which includes two working stages: traffic sign detection and traffic sign recognition. A robust shape and colour-based approach is adapted for the detection of traffic signs. Image processing techniques such as image thresholding, Gaussian filter, contour detection and RGB color segmentation have been applied for traffic sign detection. The detected traffic signs are classified based on the shape and color properties of the Indian traffic signs and recognized by region of interest (ROI) segmentation based on feature matching. The developed technique is efficient in detecting and recognizing traffic signs with complex natural backgrounds under various lighting conditions and requires a low processing time of 0.30 s, allowing for real-time applications.