Research On Vehicle License Plate Location And Tracking Technology Base On DSP

Intelligent transport surveillance system, which is closely related to people’s daily travel, plays an important role in constructing a smart city. Currently surveillance system using standard definition image based on the PC platform is unable to monitor targets with high precision, furthermore, PC’s processor is expensive, far from manageable and portable. However, high definition image surveillance system based on DSP platform can achieve accurate target monitoring by taking full advantage of pixels information in high definition image. Therefore, it’s trend to research and develop surveillance system using high definition image based on DSP platform. The technology of detecting and tracking vehicle license plate in this paper, not only works with high definition image, but also tracks vehicle license plate instead of vehicle, which breaks the classic pattern, hence introducing some new problems to be solved, such as the accuracy of license plate location and tracking, traceability range of license plate, etc. In order to give better play to license plate surveillance system on DSP platform, effective detecting and tracking algorithm is required.In this paper, we further research license plate location and tracking technology based on DSP in detail on the fundamental of summarization of existed studies. Our work mainly includes presenting a scheme of license plate monitoring based on DSP, analyzing the hardware’s advantage of DSP and the features of license plate, formulating a reasonable hardware framework, and implementing software algorithm. In this paper, we pick TMS320DM648as core to formulate hardware framework. The whole framework is divided into two functional modules—location module and tracking module. Furthermore, we propose a novel license plate location and tracking algorithm, firstly, locating the exact position of license plate by convolution and projection algorithm, then obtaining accurate information of license plate features, lastly, acquiring trace of the target vehicle by tracking the license plate features using Lucas-Kanade algorithm. As experiments show, in contrast to other vehicle monitoring technology, our two independent modules attempt fully demonstrated DSP system’s multi-pipeline feature. The detecting and tracking system takes vehicle license plate as monitoring target to achieve high-precision vehicle surveillance, which verifies the validity of DSP-based vehicle license plate detecting and tracking technology.