基于虚拟线圈的内河船舶流量检测方法

实时检测内河船舶流量对水上交通管理具有重要意义.为实时检测船舶流量,研究了一种基于虚拟线圈的船舶流量检测系统.虚拟线圈即在视频图像上设置一个封闭区域,根据该区域内图像的变化检测是否有运动目标通过.利用RGB三通道背景差分法得到视频图像的二值化图像,二值化图像的三个分割阈值由大津法求出.设置2个平行的虚拟线圈,通过虚拟线圈的船舶会被检测并计数,同时检测船舶的船长与船宽,利用BP神经网络对船舶进行分类.通过在武汉长江大桥和武汉长江二桥上不同时间段采集的视频进行实验,结果表明,船舶计数正确率达到97.1%,计数漏检率2.9%,计数错检率0%,船舶分类正确率98.6%.处理一帧图片的平均时间为7 ms,具有较好的实时性. 

A Method for Detection of Ship Traffic in Inland Waterways Based on Virtual Loop

Real time detection of ship traffic flow in inland waterways is of great significance to maritime transportation management.In order to detect ship traffic flow in a real-time mode, a detection system based on virtual loop is presented in this paper.The virtual loop is a closed area within a video image, which can detect moving targets according to the change of the image in the regions.Firstly, a binary image is obtained by RGB three-channel background difference method, in which three segmentation thresholds are obtained by Otsu method.Secondly, the ships that traverse the virtual loops are detected and recorded by two paralleled virtual loops.Besides, the length and width of the ships can also be detected.Finally, the detected ships are further classified by BP neural network.The videos collected from Wuhan Yangtze River Bridge and Second Wuhan Yangtze River Bridge at different time periods are selected as a case study.The results of the case study show that the success rate of detecting the ships reaches 97.1%, the missing rate is 2.9%, and the false alarm rate is 0%.The accuracy of classification is 98.6%.In addition, the proposed method also reveals that the average time of processing one frame is 7 ms, which indicates a good performance in term of timeliness.