【摘要】：目前,無人機(jī)已被應(yīng)用到交通管理與控制領(lǐng)域中,成為傳統(tǒng)交通監(jiān)控技術(shù)的一種有效的輔助和補(bǔ)充,并且無人機(jī)車輛檢測(cè)和跟蹤已成為該領(lǐng)域的研究熱點(diǎn)。但是,基于無人機(jī)的車輛檢測(cè)和跟蹤還存在一些問題。首先,無人機(jī)從高空拍攝的道路監(jiān)控信息容易受到車輛的運(yùn)動(dòng)、建筑物和樹木的陰影以及道路連通空白區(qū)域的光差、天氣條件等外界自然因素的干擾而出現(xiàn)誤判;另一方面,由于無人機(jī)搭載的攝像機(jī)可能做旋轉(zhuǎn),移動(dòng)和滾動(dòng)等動(dòng)作,致使無人機(jī)監(jiān)控平臺(tái)的攝像機(jī)會(huì)發(fā)生頻繁變化,從而會(huì)影響車輛跟蹤效果。本文為有效解決無人機(jī)道路監(jiān)控與車輛跟蹤方面存在的問題,提出了基于無人機(jī)航拍視頻的車輛檢測(cè)和跟蹤算法。共分為三個(gè)部分:航拍圖像道路提取、基于無人機(jī)拍攝視頻的車輛檢測(cè)以及車輛跟蹤。本文主要貢獻(xiàn)如下:(1)針對(duì)航拍圖像道路檢測(cè)易受車輛、建筑物、陰影遮擋以及道路連通空白區(qū)域的光差、天氣等外界自然因素干擾等問題,提出了一種多方法融合的道路提取算法。該算法首先根據(jù)建立的顏色模型,應(yīng)用圖像分析技術(shù)分析道路的連接特性和寬度特征;然后,運(yùn)用Hough變換提取圖像中道路像素;使用交集處理方法去除圖像中的噪聲;最后通過道路陰影顏色分析,噪聲分類處理以及道路修復(fù)等技術(shù),快速高效地從復(fù)雜的航拍圖像中提取出道路。(2)提出了基于UAV收集的圖像數(shù)據(jù)的新型車輛檢測(cè)跟蹤系統(tǒng)。主要包括四個(gè)模塊:圖像配準(zhǔn),圖像特征提取,車輛形狀檢測(cè)和車輛跟蹤。在連續(xù)圖像中引入多個(gè)車輛特征點(diǎn)來檢測(cè)車輛,以提高車輛檢測(cè)和跟蹤的系統(tǒng)準(zhǔn)確性。現(xiàn)場(chǎng)測(cè)試表明,在不同拍攝高度本系統(tǒng)對(duì)交通信息采集的精度較高,可用于未來市區(qū)的交通監(jiān)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.
【學(xué)位授予單位】：山東理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：U495

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 曾友州;胡瑩;曾偉一;鄭曉霞;;提取數(shù)字圖像邊緣的算法比較[J];成都航空職業(yè)技術(shù)學(xué)院學(xué)報(bào);2009年04期

2 宋建中;;噴霧圖像的自動(dòng)分析[J];光學(xué)機(jī)械;1988年04期

3 涂承媛;曾衍鈞;;醫(yī)學(xué)圖像邊緣快速檢測(cè)的模糊集方法[J];北京工業(yè)大學(xué)學(xué)報(bào);2005年06期

4 常君明;馮，

本文編號(hào)：2219183