Detection and recognition of traffic signs based on RGB to red conversion

This paper presents a different approach for detecting and recognizing traffic signs. Traffic signs are designed such that they can be easily observed by humans because of their shape and shading which are not precisely the same as surrounding environment. Traffic sign recognition is a fundamental component in autonomous cars and it can similarly be utilized as an integral part of driver assistance frameworks. The system is designed to perform two tasks, traffic sign detection and recognition. The algorithms demonstrated include RGB to Red conversion for getting the red component in the image, filters which perform the task of noise reduction as well as edge detection, thresholding and segmentation to get the desired traffic sign from image and then an artificial neural network for recognition. The neural network used here is a Single Layer Perceptron neural network. Implementation of the system has been successfully done in Matlab and it works quite productively and efficiently with the database of Indian traffic signs created as a part of the project.