Text extraction algorithm for traffic signs
Zhiqiang Wei · Journal of Computer Applications · 2011
A fast and robust approach for the text extraction of traffic signs based on color and stroke was proposed.First,a new color model derived from Karhunen-Loeve(KL) transform was applied to find all possible traffic sign candidates.Then,affine transformation was performed to restore traffic signs to let every road sign seem to be vertical to the camera optical axis which can improve the accuracy in detecting texts embedded in traffic signs.Finally,mathematical morphology and region growing algorithms were used to obtain a clearer binary picture which was sent to Optical Character Recognition(OCR) software.The experimental results demonstrate great robustness and efficiency of the proposed algorithm.