Research And Application Of Vehicle License Plate Recognition Based On ARM9

In recent years,the number of motor vehicles in our country has maintained a high growth trend.The increasing number of cars has brought about a series of problems,such as traffic jam,traffic accidents and so on,while convenient for people’s life.Traditional manual management is inefficient.Intelligent transportation system integrated with information,computer and image processing technology is the main trend of future transportation system.As an important part of intelligent transportation system,license plate recognition system is one of the hot research topics in the field of high technology.Based on the research of embedded system and license plate recognition algorithm,this manuscript proposes an embedded license plate automatic recognition system based on ARM,aiming at the disadvantages of large volume and difficult installation of PC machine license plate recognition system.The system is mainly divided into two parts: hardware and software.The specific work of the hardware part is as follows:(1)Design hardware system.The hardware system mainly includes ARM9 chip S3C2440,power module,memory module,camera module,network and interface module.(2)Build an embedded Linux platform,including the modification and transplantation of BootLoader programs,the compilation and transplantation of the Linux kernel,the production and transplantation of the root file system and so on.The software part is divided into three stages: license plate location,character segmentation and character recognition.The specific work is as follows:(1)Propose a positioning method combining the multiple license plate location method with the license plate judgment.First,the first location of the license plate location method based on HSV color space is used for the first time;then,the edge detection method based on the Sobel operator is used for two positioning;finally,the third license plate location is performed by the method extracted by the maximum stable extremum region(maximally stably extremal regions,Mser).In this manuscript,Support Vector Machine(SVM)license plate judgement model is trained by license plate location.By setting the threshold of license plate detection,different application scenarios can be satisfied.(2)In order to improve the accuracy of the license plate character segmentation,this manuscript uses the gray jump principle to remove the border and rivets before the license plate segmentation,and then uses the vertical projection to divide the license plate with the prior information of the license plate of our country,and obtains the independent license plate character.(3)In order to improve the accuracy and real-time of the license plate recognition system,a license plate recognition algorithm based on improved Local Binary Patterns(LBP)and Multilayer Perceptron(MLP)is proposed in this manuscript.Aiming at the problem that LBP operator can not describe the location information of characters,and Chinese characters are more complex than digital and alphabetic texture.In this manuscript,we divide the normalized Chinese characters,letters and numbers into different blocks.And then get the feature vectors of the characters by combining the LBP feature extracted from each small block.The chinese characters MLP model and the alphanumeric MLP model are trained by the extracted character eigenvectors.In order to improve the recognition accuracy,the MLP model is trained by error back propagation.The test results show that the accuracy of the character recognition method proposed in this manuscript can reach over 98.5%.For the realization of the embedded license plate automatic recognition system,this manuscript developed a license plate automatic recognition software based on OpenCV,and successfully compiled and run on the PC machine.Finally,the license plate recognition and the OpenCV library were cross compiled and transplanted to the circuit board.