基于快速背景建模的垃圾车识别系统

Garbage Truck Identification System Based on Fast Background Modeling

为保障城市道路交通安全、实现绿色环保出行,城市道路智能监控愈发重要,重点监控车辆如城市垃圾车的智能识别也成为其中关键一环。为提高城市道路智能监控系统分析能力,设计了基于快速背景建模的垃圾车车型识别系统。首先,设计了垃圾车识别系统的网络拓扑结构,并构建了车辆视频侧视采集系统以保留丰富的车辆特征。其次,为降低复杂背景的影响,提出背景更新控制策略,实现了快速可靠的背景建模;为实现侧视车辆图像的有效分割,提出了基于动态位置检测的关键帧筛选方法和基于斜线投影的目标分割方法。最后,设计了基于SVM(Support Vector Machine)的多级分类器实现了垃圾车识别。试验结果表明,文中方法垃圾车识别的平均准确率为95%以上、平均召回率为90%以上。

In order to ensure the safety of urban road traffic and meet the green travel,intelligent monitoring of urban roads is becoming more and more important,and the intelligent identification of key monitoring vehicles such as urban garbage trucks has also become a key link.In order to improve the analysis ability of urban road intelligent monitoring system,a garbage truck model identification system based on fast background modeling is designed.Firstly,the network topology of garbage truck identification system is designed,and the vehicle video side view acquisition system is constructed to retain rich vehicle characteristics.Secondly,in order to reduce the influence of complex background,the background update control strategy is proposed to realize fast and reliable background modeling,and to realize the effective segmentation of side-view vehicle images.A key frame filtering method based on dynamic position detection and a target segmentation method based on oblique projection are proposed.Finally,a multi-level classifier based on SVM is designed to realize garbage truck recognition.The experimental results show that the average accuracy of garbage truck identification is over 95%and the average recall rate is over 90%.