Development Of The Vehicle Plate Recognition System Based On Mathematical Morphology

With the rapid growth of the automobile enterprise and transpotation, the world requires the intelligent management of traffic management. License Plate Recognition system (LPR) is the core of Intelligent Traffic System (ITS). It is the main part of the traffic information service system and plays an important role in the intelligent control system.The license plate recognition is composed of three steps, i.e., license plate location, character segmentation, and character recognition. Although mathematical morphology has been employed in license plate location, it has not been used in other two steps. In this paper, mathematical morphology and other techniques are used in license plate location and character segmentation to improve their performance and the license plate recognition rate is therefore significantly increased. At the stage of license plate location, Morphology and Topology are employed in plate location. Firstly, image preprocessing is adopted, which is implemented by Top-Hat transform, opening and closing operations. Secondly, connected components analysis is performed to detect candidate regions. At last, the real license plate area is obtained according to the Euler Number of candidate regions. During the stage of character segmentation, improved projection algorithm based on Morphology and Projection is used in segmentation. Firstly, morphology is adopted to intensify the whole characters area. Secondly, horizontal projection is performed to eliminate horizontal plate frames and vertical projection is used to intensify the single character area. At last, every character is segmented according to the characteristic features of the plate. In the character recognition stage, template matching method is used in recognition. Characters should be normalized at first. And then we abstract the characters'thread attribute, periphery attribute, and formed a vector, and compare it with the attribute vectors in character template library, so the characters are recognised. This LPR algorithm based on Mathematical Morphology has good accuracy and time effectiveness.At last, a hardware of license plate recognition system based on FPGA is designed by SOPC, and the software design of the system is accomplished by Nios II soft core processor.