Image Based Detection and Recognition of Road Signs

The objective of our work is to develop a noise tolerable threshold based segmentation and detection , which is learnt to the rotation, position and scaling also. Traffic sign analysis can be divided into three main problems: Find location, Detection and Recognition of traffic signs. In existing approaches locating and detection of traffic sign is based on color information and shape based extraction. In this paper, the original traffic image is converted into a grayscale image. The gray scale image was performed into Sobel edge detection algorithm to detect the edges. Filling the interior gaps, which is used to avoid the noise in an image. Smoothen of the object is done using the erosion operations. Segmentation technique is used to remove the unused area of the traffic signs. The clear traffic symbol and fast accessing were obtained from our method. After that the features are extracted from segmented image using histogram technique. SVM classifiers are used in the recognition process to classify and display the processed output image. The advantages of our approach are simple, fast and able to identify a common road sign.