多车道的车流量统计  

Statistics of the Traffic Flow on Multiple Lanes

作　　者：李宇成[1] 李雷[1] 王目树[1] 

机构地区：[1]北方工业大学机电工程学院,北京100144

出　　处：《自动化仪表》2010年第10期57-60,共4页Process Automation Instrumentation

摘　　要：针对视频车流量检测容易受光线变化以及相邻车辆阴影影响的问题,提出一种基于车辆图像能量差分的车流量检测方法。该方法利用车辆视频图像的能量信息作为车辆的检测特征,辨别检测区域各部分状态,减弱光照、阴影等的影响;并且通过实时自动提取和更新背景能量图的方式来检测车流量。实验表明,与传统的背景帧差法相比较,该方法受光照变化、车辆阴影的影响较小,实时性好、准确度高。The video traffic flow detection is easily affected by changes of the light and the shadows of adjacent vehicle.To solve these problems,the new detection method based on image energy difference is presented.By using the energy information of video image of vehicle as the detecting characteristics of vehicle,the status of each detection area is identified,and the influence of intensity of the light and shadows are reduced.In addition,the background energy image is automatically extracted and updated in real-time for detecting traffic flow.The experiments indicate that comparing with traditional background difference,the new method is better in real-time performance with higher accuracy and less influence to light intensity and shadows.

关 键 词：车流量检测 图像能量 智能交通 视频检测 图像识别 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]