Abstract:
Detection and classification of traffic sign are essential module of autonomous driver assistance
systems(ADAS) playing a significant role to the safety of pedestrian, vehicles specially drivers but the
problem remain still challenging because most of the drivers are less educated and don’t know the
meaning of many traffic signs which can cause serious problems. We present an active and capable
system for detection and classification of traffic signs name “Traffic Sign Recognition (TSR) with voice
notification”. It is an applicationsystem which assists user by detecting and recognizing traffic sign from
given input image and alert the user through its voice notification feature. In this application, we focus
more on voice feature because the main objective of this application is to provide meaningful voice
notification. Detection of traffic sign in candidate image is achieved by conversion of RGB image to HSV
color space and after color space conversion, a color based division has been applied. Geometrical
constraints are applied to the interested regions after segmentation. The final interested region is extracted
through applying Hough transform to the segmented image for searching triangle, circle or a rectangle in
candidate image. if sign is detected in input image, recognition or classification methods are carried out
by using most famous feature matching technique name Speed Up Robust Features (SURF). After
successful recognition of sign, we perform a text to voice conversion method which give a meaningful
voice notification to user. The selected methodology results checked on a data set of 15 most frequently
occurring traffic signs reported good results on providing right voice notification.
