The Algorithm Research And The Android System Design Based On The License Plate Recognition

With the economic development in our country, the number of motor vehicles is also increasing sharply. So an intelligent management of transportation to decrease the traffic jams has become an important issue. As a key technology for intelligent transportation, license plate recognition has been developed for many years. It not only can be used to manage and control traffic, and can also be used in residential areas, schools, underground car parks and corporate park. But in practice, the recognition technology of license plate has still not been extended because of its limitations of research on vehicle license plate recognition. Therefore, it is worth for us to explore this issue.Based on digital image procession technology, computer vision technology and artificial neural network technology, the paper deeply researches and analyses an automobile license plate identification system. And I have finished following works in this article:(1) This paper proposed compressed sensing (Compressed Sensing, CS) method for de-noising. Using the proposed method of image preprocessing can effectively recovery the original image information of the license plate. We take the noise problem into account. This is because that the actual picture of cars is susceptible to noise interference. According to this fact, the preprocessing of license plate image is necessary. Removing the license plate image noise is its main purpose. Therefore,(2) In the license plate locating, in order to avoid uneven lighting, brightness and so on, we just handle saturation layer. In addition, we also introduced an adaptive Wiener filter. It can cause the background image blurring for its excessive noising. For this reason, we can extract the original license plate area of the image, and get the candidate region of the license plate by the projection method. Due to China’s automobile license plate aspect ratio is fixed, we can get license plate from the candidate area with the closest ratio.(3) In the license plate segmentation, the preprocessing of license plates is similar to vehicle license plate locating. We can get the column-oriented pixels of the license plate by projection method, and then we can get the character by difference between pixels of character and space.(4) In the character recognition, we used matching method of Haursdorff distance from edge. In experiments, we can demonstrate that the proposed method of the character recognition has the higher rate and the higher operation speed.(5) In this paper, the Android build environment is built by Eclipse and Android SDK(Software Development Kit, SDK). It integrates findings above. At the same time, to achieve the vehicle’s location is base the Gaode map development package. With the development of smart phones, it occupies an important position not only in people’s daily life but also in their work. So the development of mobile’s license plate recognition system is necessary. It is convenient for polices and law enforcement officers.