Design And Implementation Of FPGA Architecture For License Plate Detection Based On Sliding Window

In license plate identification system,correct locating of plates contributes to realization of the whole system's function. Its accuracy is related to the efficiency and performance of following plate divided identification. However, it is difficult to fully achieve that for license plate identification systems which exist as software application so far.In this paper, design and realization of the FPGA structure plate identification based on sliding window algorithm arepresented with the advantage of parallel processing in FPGA structure. Firstly, adopting windows in fixed scale, slide them without overlapping on images of license plates and obtain candidate region through calculation. Then, in the candidate region, use a multi-scale window for dynamic slide and achieve final locating according to geometrical features.Finally, license plates abstracting is realised with small sacle sliding window and HSI filter, and division and identification of plates are follwed.Stimulate and vertify with Modelsim, and test results applying DE2-115 hardware platform.The result shows : adopting sliding window plate algorithm which is achieved with FPGA hardware structure, real-time identification of 800?600 resolution video image is realized at 50 MHz of system clock.Meanwhile, it ensures following divided identification stability.