实时车辆排队长度图像检测系统及FPGA硬件实现  

Design and FPGA Implementation of Real-Time Vehicle Queue Detection System Based on Image Processing

作　　者：王闯[1] 史忠科[1] 

机构地区：[1]西北工业大学自动化学院,西安710129

出　　处：《交通运输系统工程与信息》2012年第3期65-74,97,共11页Journal of Transportation Systems Engineering and Information Technology

基　　金：国家自然科学基金重点项目(61134004)

摘　　要：针对现有交通路口车辆排队长度检测系统处理速度低、无法适用于实际复杂路口环境的缺陷,设计了一种基于FPGA的实时车辆排队长度图像检测系统.通过形态学边缘处理检测车辆的存在,对检测出的车队轮廓进行水平投影并采用一种基于信息量的车辆度量方法从投影结果中提取了排队的队尾,同时在确定排队长度时通过对算法参数的自适应调整有效消除图像逆透视效果的干扰;整个图像处理过程完全利用FPGA硬件逻辑实现,采用五级并行流水设计保证图像采集、高斯滤波、Sobel边缘检测、阈值分割和形态学腐蚀处理同步执行且同时完成,实现了对分辨率为720×576的图像25帧/s的处理速度.实验结果表明,系统性能良好,工作稳定,算法简单实用,具有较好的应用前景.In view of the defects of existing vehicle queue length detection systems in processing speed and adaptability to the real complex intersection environment, the paper proposes a FPGA-based real-time vehicle queue detection system. In this system, morphological edge detection is used to detect the vehicle presence, and the horizontal projection processing is applied. Next, the entropy-based vehicle measurement method is adopted to detect the rear of the queue. Interference of the perspective effect of image sequence is eliminated through adaptive algorithm parameters adjustment. The hardware scheme of FPGA-based multilayered image processing possesses a linear five-level pipeline architecture, which makes the operation of image acquisition, gauss filtering, Sobel edge detection, threshold segmentation and morphological erosion can be completed almost at the same time. The system achieves up to the real-time processing speed of 25 frames per second for 720 × 576 pixel grayscale images. Extensive field experiments show that the proposed detection algorithm is simple and practical, the system is stable and promising.

关 键 词：交通工程 车辆排队长度 FPGA 实时图像处理 形态学边缘处理 

分 类 号：U491[交通运输工程—交通运输规划与管理]