车牌定位算法的硬件实现

Hardware Accelerator of Car License Plate Location Algorithms

车牌定位技术是整个车牌识别系统的关键技术,也是最难点技术,准确的车牌定位算法有助于识别系统后续工作的顺利进行。通过对车牌定位算法的深入理解和研究,从车牌定位的准确性、实时性、经济性三个方面考虑,设计了一种基于灰度图像的实时定位算法。算法主要包括对车牌的灰度图像进行水平差分、边缘检测和数学形态学运算,最后完成车牌区域的正确查找。为了进一步提高算法的实时性,利用Altera开发的DE2-115 FPGA为开发板,搭建了定位算法的各个模块硬件电路。实验结果表明,该算法在车牌定位的领域具有良好的性能。

The car license plate location technique is the key point and the most dithcult part m the license plate recognition system. An accurate car license location is helpful for recognition system. Through in-depth understanding and research on the algorithms of license plate location, this paper describes the design of a real-time positioning algorithm based on gray image considering the accuracy, real-time performance and the cost. The algorithm mainly includes horizontal difference on the red channel, edge detection, and mathematical morphology operations, and then it finds the license plate region correctly. In order to improve the real-time performance, hardware circuit of each module is built on the DE2-115 FPGA developing board. The experimental results show that the algorithm has a good performance in the field of car license plate location.