Vision Based Road Sign Detection and Recognition

Traffic regulation is an important concept in the modern society. For maintaining safety in the flow of traffic, government all over the world have taken steps to develop certain rules, known as traffic rules. Some of the traffic rules are directly displayed on roads in terms of traffic signs. Traffic signs are those that use a visual/symbolic language about the road ahead that can be interpreted by drivers for making driving safe and convenient. This paper proposes an autonomous traffic sign detection, which would detect, recognize and interpret the meaning of the traffic signs for the driver and will provide a great help in reducing the road accidents and the deaths caused by it. The system mainly has two phases, 1) Sign detection, 2) Recognition. In detection phase a two stage algorithm is performed (i) detection through color image segmentation using HSV color space on color-images (ii) Validation of the obtained region of interest taking advantage of the shape properties of the road signs. In recognition phase the regions of interest are finally represented using HOG descriptors and are fed into trained Support Vector Machines (SVMs) in order to be classified and finally the driver is alerted by providing voice indication of the recognized sign.