Detekce dopravních značek z obrazů kamer zabudovaných ve vozidlech (trojúhelníkové značky)

Abstract

The aim of this work was to propose and implement an algorithm for detection of triangular traffic signs in an image. This kind of algorithm can be used not only for traffic signs condition control, but also as a driver assistant system improving traffic safety. Color segmentation based on each color channel thresholding is used for localization of the sign in an image. Artificial neural network classifies localized traffic signs. Using two methods mentioned above should ensure the speed empowering real-time using. Implemented application allows setting up necessary parameters and provides the results of image analysis as well as analysis progress information to user.