A Hybrid Approach for Detection and Recognition of Traffic Text Sign using MSER and OCR

Abstract

Detection and Recognition in traffic sign pictures or common pictures has applications in Computer vision frameworks like enlistment number plate identification, programmed movement sign location, picture recovery and help for outwardly disabled individuals. In this paper a hybrid approach based on MSER and OCR, utilizing clamor expulsion strategy, i.e. Lucy-Richardson calculation. After clamor evacuation, content district location stage begins with complexity upgraded edge improved MSER area discovery system is utilized there after morphological division is utilized to section content locale in the picture. After location stage acknowledgment stage begins in which content applicants are separated utilizing geometric filtration utilizing properties, for example, viewpoint proportion, unusualness, solidicity, and so forth. At that point Bounding box strategy is utilized to distinguish letter competitors and shape word out of them. At long last, Optical Character Recognition (OCR) instrument is utilized to concentrate message out of picture. The framework displayed beats best in class techniques on the dataset of the movement content sign information that were gotten from Jaguar Land Rover Research.