第40卷第3期 2017年9月
上海船舶运输科学研究所学报
JOURNALOF SHANGHAI SHIP AND SHIPPINGRESEARCH INSTITUTE
文章编号：1674-5949(2017)03-0064-07
基于视频的双模型卡口触发算法
刘俊，谭中慧
（中远海运科技股份有限公司，上海200135）
Vol. 40 No. 3 Sep. 2017
摘要：地磁感应线是一种传统的卡口触发信号源，该信号源原理筒单，性能稳定可靠，但其施工不便、容易损毁等诸多间题在现代交通中逐渐凸显。随着视频监控的高清化、数字化程度逐渐提高及数字图像处理技术的进步出现了以视频图像本身为信号源的牌照识别卡口触发和基于多级分类的车辆模型触发等新的卡口触发方式。前者对于车牌污损和无牌车辆不能提供较好的解决方案.后者不仅计算量大·且较难提取出通用性较好的车辆分类器。现提出利用视频图像虚拟线圈技术.利用视频灰度时序实现车辆目标触发，并通过车辆平面模型编码特性消
除遮挡误差。该算法运算速度快，可自适应地动态调整检测参数，具有较高的环境适应性和鲁棒性等优点。关键词：卡口；线圈；视顿；灰度；遮挡
中图分类号：TP391.41
文献标志码：A
VideoBasedDualModeTrafficTriggeringAlgorithm
LIUJun.TANZhonghui
(C(SCO)SHIPPING TechnologyCo.,Ltd.,Shanghai 200135,China)
Abstract: Geonagnetie induction coil is a traditional traffic trigger signal source, The principle is straightforward, and the per formance stable and reliable, But it is easy to damage and involves construction work. On the other hand, as the improvement of the video surveillance equipment and the progress of the digital image processing technology, some new traffic trigger modes based on video itself, such ax video license plate recognition and multi-level classification of vehicle model triggering, come a-long. The former can not deal with license plate defacement or unlicensed vehicles, and the latter is not only computationally in-tensive. but vehicle classification model dependent, A meaningful vehicle classification model takes effort to build, This paper proposes the video image virtual coil technology. with the video gray level timing to achieve vehicle target trigger, and the vehi cle discrimination on basis of their plane model coding characteristics to eliminate the ocelusion error, The algorithm is fast and
adaptive, and hax high ambient tolerance and robustness. Key words; traffic; detection coil; video: grayscale; shelten
0引言
交通卡口抓拍系统在国内推行时间较早，该系统具有广泛用途。传统卡口抓拍系统的设备主要以一种纯物理接触式的地磁感线圈触发为主，该系统原理简单、使用时稳定可靠。但线圈铺设需开挖路基，影响交通，而且长时间受到车辆碾压容易损毁。近年来，随着交通压力的逐步加大，线圈的弊端逐渐显露。
照识别触发是一种视频有源触发，将牌照作为目标源，在图像目标区域内不断进行扫描、搜索，定位牌照从而实现精确触发，但牌照识别触发方式对于牌照污损的车辆和无牌车辆不能提供很好的解决方案。Haar分类
收稿日期：2017-06-05
作者简介：刘俊（1977一）.男，江西部阳人·高级工程师，主要从事智能电子产品研究。万方数据