發(fā)布時間：2018-01-13 17:30

  本文關(guān)鍵詞：高速公路惡劣天氣及交通狀況智能分析系統(tǒng)研究與實(shí)現(xiàn)　出處：《西南交通大學(xué)》2014年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 高速公路 視頻監(jiān)控 惡劣天氣 車輛跟蹤 運(yùn)動檢測

【摘要】：惡劣的天氣及交通異常狀況常常給高速公路的安全帶來巨大威脅,高速公路管理部門需要實(shí)時地掌握其轄區(qū)內(nèi)各個路段的氣象情況及交通狀況,以便做出及時準(zhǔn)確的應(yīng)對決策。然而,目前主要依靠傳感器等硬件設(shè)備提供的數(shù)據(jù),經(jīng)分析得到氣象信息和交通信息,安裝及維護(hù)成本高昂,且使用時間和環(huán)境條件可能影響其準(zhǔn)確性。 本文充分利用已經(jīng)建立的高速公路監(jiān)控體系,研究了基于高速公路監(jiān)控視頻的惡劣天氣(霧、雨、雪)檢測和交通異常狀況(擁堵、停車、逆行)檢測算法,通過對監(jiān)控攝像頭傳入的視頻數(shù)據(jù)的智能分析,得到該路段的天氣信息和交通信息。本文算法結(jié)合了圖像處理、計算機(jī)視覺及模式識別等領(lǐng)域的先進(jìn)技術(shù),對視頻圖像中的霧、雨、雪的視覺特征進(jìn)行提取和分析,從而實(shí)現(xiàn)惡劣天氣的檢測和報警,同時對運(yùn)動車輛進(jìn)行跟蹤,并對其進(jìn)行行為分析,從而實(shí)現(xiàn)交通異常狀況的檢測和報警。 本文分析了霧、雨、雪在監(jiān)控視頻中呈現(xiàn)出的視覺特征,針對霧區(qū)的模糊效應(yīng)提出了基于Canny邊緣的霧天檢測算法。對于雨和雪的檢測,基于監(jiān)控視頻分辨率小及噪聲干擾大的特點(diǎn),本文算法放棄傳統(tǒng)的對雨滴和雪花的動態(tài)特性進(jìn)行分析的方法,分析了雨天和雪天在車道上的視覺特征,針對濕滑道路的反光特性和圖像鈍化的特征,提出了反光度與圖像銳度結(jié)合的道路濕滑程度評估方法,從檢測道路是否濕滑的角度來實(shí)現(xiàn)雨天的檢測,針對道路上積雪的顏色特征,提出了基于積雪顏色模型的積雪檢測方法,從檢測道路上是否覆蓋積雪的角度來實(shí)現(xiàn)雪天的檢測。 針對交通異常狀況的檢測,本文首先分析了前景檢測和背景重建過程中幾種常用的技術(shù),并通過實(shí)驗(yàn)對其效果進(jìn)行對比,經(jīng)分析,本文算法采用背景差分法與基于統(tǒng)計模型的背景重建與更新方法相結(jié)合,實(shí)現(xiàn)運(yùn)動車輛的檢測和跟蹤,估算出車速、車流量及車輛行駛方向,并運(yùn)用這些信息進(jìn)行綜合分析和判斷,實(shí)現(xiàn)高速停車、逆行及擁堵狀況的檢測。 本文使用了豐富的視頻數(shù)據(jù)作為實(shí)驗(yàn)數(shù)據(jù)集,通過對實(shí)驗(yàn)結(jié)果數(shù)據(jù)的統(tǒng)計和分析,說明了本文算法在惡劣天氣檢測及交通異常狀況分析中的有效性�；诒疚乃惴▽�(shí)現(xiàn)了具有實(shí)用價值的智能分析系統(tǒng),該系統(tǒng)與視頻監(jiān)控平臺兼容,不僅能分析本地歷史視頻,而且能以多路并行輪詢的方式分析在線實(shí)時視頻,是高速公路信息服務(wù)體系中的重要組成部分,功能完整,靈活性高,它的應(yīng)用將有效提高高速公路的管理能力和應(yīng)急能力。
[Abstract]:The bad weather and abnormal traffic situation often bring great threat to the safety of freeway. The highway management department needs to grasp the meteorological situation and traffic condition of each section in its jurisdiction in real time. In order to make timely and accurate response to the decision. However, at present, mainly depends on the sensor and other hardware equipment to provide data, through analysis to obtain meteorological information and traffic information, installation and maintenance costs are high. And the use time and the environment condition may affect its accuracy. In this paper, we make full use of the established highway monitoring system, and study the detection algorithm of bad weather (fog, rain, snow) and traffic anomaly (congestion, parking, retrograde) based on highway surveillance video. Through the intelligent analysis of the video data from the surveillance camera, the weather information and traffic information of the section are obtained. This algorithm combines the advanced technology of image processing, computer vision and pattern recognition. The visual features of fog, rain and snow in video images are extracted and analyzed to detect and alarm bad weather, and track and analyze the behavior of moving vehicles. In order to realize the traffic abnormal condition detection and alarm. In this paper, the visual features of fog, rain and snow in surveillance video are analyzed. Aiming at the fuzzy effect of fog region, a fog detection algorithm based on Canny edge is proposed. Based on the characteristics of small resolution and large noise interference, this algorithm gives up the traditional method to analyze the dynamic characteristics of raindrops and snowflakes, and analyzes the visual features of rainy and snowy days in driveway. According to the reflective characteristics of wet slippery road and the feature of image passivation, this paper proposes a method to evaluate the wet slippery degree of road, which combines reflectance and image sharpness, and realizes the detection of rainy day from the angle of detecting whether the road is slippery or not. According to the color characteristics of snow cover on the road, a snow detection method based on snow color model is proposed to detect snow cover on the road. For the detection of traffic anomaly, this paper first analyzes several common techniques in the process of foreground detection and background reconstruction, and compares their effects through experiments. In this paper, the background difference method is combined with the background reconstruction and update method based on statistical model to realize the detection and tracking of moving vehicles, and to estimate the speed, flow and direction of vehicles. This information is used for comprehensive analysis and judgment to detect high speed parking, retrograde and congestion. In this paper, we use rich video data as experimental data set, through the statistics and analysis of experimental data. The effectiveness of this algorithm in the detection of severe weather and the analysis of traffic anomalies is explained. Based on this algorithm, an intelligent analysis system with practical value is implemented, which is compatible with the video surveillance platform. Not only can analyze the local history video, but also can analyze the online real-time video in the way of multi-channel parallel polling. It is an important part of the expressway information service system, with complete function and high flexibility. Its application will effectively improve the ability of expressway management and emergency response.
【學(xué)位授予單位】：西南交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：U495;TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 許亮;賈振紅;覃錫忠;;視頻圖像中雪花的檢測和去除[J];光電子.激光;2007年04期

2 段瑞玲,李慶祥,李玉和;圖像邊緣檢測方法研究綜述[J];光學(xué)技術(shù);2005年03期

3 林洪文,涂丹,李國輝;基于統(tǒng)計背景模型的運(yùn)動目標(biāo)檢測方法[J];計算機(jī)工程;2003年16期

4 曹江中;戴青云;譚志標(biāo);邸磊;;基于視頻的高速公路車輛檢測和跟蹤算法[J];計算機(jī)應(yīng)用;2006年02期

5 張怡;張加萬;楊國強(qiáng);李亮;;基于時空一致性優(yōu)化的視頻去霧算法[J];計算機(jī)應(yīng)用研究;2011年10期

6 石茂清;惡劣氣象條件下的道路交通安全研究[J];交通標(biāo)準(zhǔn)化;2005年08期

7 孫洪運(yùn);楊金順;李林波;吳兵;;惡劣天氣事件對道路交通系統(tǒng)影響的研究綜述[J];交通信息與安全;2012年06期

8 王柏林;胡玉峰;李佳;;移動式自動氣象站設(shè)計及其在應(yīng)急氣象環(huán)境監(jiān)測服務(wù)中的應(yīng)用[J];氣象科技;2006年05期

9 田艷;張志強(qiáng);張景涵;趙玉蘭;;高速公路交通氣象監(jiān)測系統(tǒng)的研究與應(yīng)用[J];氣象水文海洋儀器;2011年02期

10 張朝林;張利娜;程叢蘭;王必正;;高速公路氣象預(yù)報系統(tǒng)研究現(xiàn)狀與未來趨勢[J];熱帶氣象學(xué)報;2007年06期

相關(guān)博士學(xué)位論文 前2條

1 嚴(yán)捷豐;交通視頻監(jiān)控中的車輛檢測與分割方法研究[D];中國科學(xué)技術(shù)大學(xué);2008年

2 許潔瓊;基于視頻圖像處理的車輛檢測與跟蹤方法研究[D];中國海洋大學(xué);2012年

，

本文編號：1419836