基于航拍图像的道路交通监控方法研究
发布时间：2018-09-03 06:30
【摘要】：目前,无人机已被应用到交通管理与控制领域中,成为传统交通监控技术的一种有效的辅助和补充,并且无人机车辆检测和跟踪已成为该领域的研究热点。但是,基于无人机的车辆检测和跟踪还存在一些问题。首先,无人机从高空拍摄的道路监控信息容易受到车辆的运动、建筑物和树木的阴影以及道路连通空白区域的光差、天气条件等外界自然因素的干扰而出现误判;另一方面,由于无人机搭载的摄像机可能做旋转,移动和滚动等动作,致使无人机监控平台的摄像机会发生频繁变化,从而会影响车辆跟踪效果。本文为有效解决无人机道路监控与车辆跟踪方面存在的问题,提出了基于无人机航拍视频的车辆检测和跟踪算法。共分为三个部分:航拍图像道路提取、基于无人机拍摄视频的车辆检测以及车辆跟踪。本文主要贡献如下:(1)针对航拍图像道路检测易受车辆、建筑物、阴影遮挡以及道路连通空白区域的光差、天气等外界自然因素干扰等问题,提出了一种多方法融合的道路提取算法。该算法首先根据建立的颜色模型,应用图像分析技术分析道路的连接特性和宽度特征;然后,运用Hough变换提取图像中道路像素;使用交集处理方法去除图像中的噪声;最后通过道路阴影颜色分析,噪声分类处理以及道路修复等技术,快速高效地从复杂的航拍图像中提取出道路。(2)提出了基于UAV收集的图像数据的新型车辆检测跟踪系统。主要包括四个模块:图像配准,图像特征提取,车辆形状检测和车辆跟踪。在连续图像中引入多个车辆特征点来检测车辆,以提高车辆检测和跟踪的系统准确性。现场测试表明,在不同拍摄高度本系统对交通信息采集的精度较高,可用于未来市区的交通监控和控制。
[Abstract]:At present, UAV has been applied to the field of traffic management and control, and has become an effective supplement to the traditional traffic monitoring technology, and UAV vehicle detection and tracking has become a research hotspot in this field. However, there are still some problems in vehicle detection and tracking based on UAV. First of all, the road monitoring information captured by UAV from high altitude is liable to be misjudged by the movement of vehicles, the shadow of buildings and trees, the light difference of the road connected to the blank area, weather conditions and other external natural factors; on the other hand, Because the camera on the UAV may rotate, move and roll, the camera of the UAV monitoring platform will change frequently, which will affect the tracking effect of the vehicle. In order to effectively solve the problems in UAV road monitoring and vehicle tracking, a vehicle detection and tracking algorithm based on UAV aerial photography video is proposed in this paper. It is divided into three parts: road extraction, vehicle detection based on UAV video and vehicle tracking. The main contributions of this paper are as follows: (1) the road detection in aerial images is vulnerable to the interference of natural factors such as vehicle, building, shadow occlusion and road connected blank area, such as light difference, weather, and so on. A multi-method fusion algorithm for road extraction is proposed. Firstly, according to the established color model, the image analysis technique is applied to analyze the road connection characteristics and width characteristics. Then, the road pixels are extracted by Hough transform, and the noise in the image is removed by the intersection processing method. Finally, the road is extracted from complex aerial images quickly and efficiently by the techniques of road shadow color analysis, noise classification and road repair. (2) A new vehicle detection and tracking system based on image data collected by UAV is proposed. It includes four modules: image registration, image feature extraction, vehicle shape detection and vehicle tracking. In order to improve the accuracy of vehicle detection and tracking, multiple vehicle feature points are introduced into continuous images to detect vehicles. The field test shows that the system has high accuracy for traffic information collection at different shooting heights and can be used for traffic monitoring and control in urban areas in the future.
【学位授予单位】：山东理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495
[Abstract]:At present, UAV has been applied to the field of traffic management and control, and has become an effective supplement to the traditional traffic monitoring technology, and UAV vehicle detection and tracking has become a research hotspot in this field. However, there are still some problems in vehicle detection and tracking based on UAV. First of all, the road monitoring information captured by UAV from high altitude is liable to be misjudged by the movement of vehicles, the shadow of buildings and trees, the light difference of the road connected to the blank area, weather conditions and other external natural factors; on the other hand, Because the camera on the UAV may rotate, move and roll, the camera of the UAV monitoring platform will change frequently, which will affect the tracking effect of the vehicle. In order to effectively solve the problems in UAV road monitoring and vehicle tracking, a vehicle detection and tracking algorithm based on UAV aerial photography video is proposed in this paper. It is divided into three parts: road extraction, vehicle detection based on UAV video and vehicle tracking. The main contributions of this paper are as follows: (1) the road detection in aerial images is vulnerable to the interference of natural factors such as vehicle, building, shadow occlusion and road connected blank area, such as light difference, weather, and so on. A multi-method fusion algorithm for road extraction is proposed. Firstly, according to the established color model, the image analysis technique is applied to analyze the road connection characteristics and width characteristics. Then, the road pixels are extracted by Hough transform, and the noise in the image is removed by the intersection processing method. Finally, the road is extracted from complex aerial images quickly and efficiently by the techniques of road shadow color analysis, noise classification and road repair. (2) A new vehicle detection and tracking system based on image data collected by UAV is proposed. It includes four modules: image registration, image feature extraction, vehicle shape detection and vehicle tracking. In order to improve the accuracy of vehicle detection and tracking, multiple vehicle feature points are introduced into continuous images to detect vehicles. The field test shows that the system has high accuracy for traffic information collection at different shooting heights and can be used for traffic monitoring and control in urban areas in the future.
【学位授予单位】：山东理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495
【相似文献】
相关期刊论文 前10条
1 曾友州;胡莹;曾伟一;郑晓霞;;提取数字图像边缘的算法比较[J];成都航空职业技术学院学报;2009年04期
2 宋建中;;喷雾图像的自动分析[J];光学机械;1988年04期
3 涂承媛;曾衍钧;;医学图像边缘快速检测的模糊集方法[J];北京工业大学学报;2005年06期
4 常君明;冯，
本文编号：2219183
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2219183.html
