基于雙目視覺的智能車道路環(huán)境識別研究
[Abstract]:With the rapid development of the automobile manufacturing industry and the general improvement of the social economic level, the per capita automobile ownership has been increasing year by year, and the automobile has become more and more popular in the world. The rapid development of automobile industry has improved the travel radius of human beings and made transportation more convenient. But the rapid development of automobile industry has also brought a lot of hidden dangers to human beings and society. For example, vehicle exhaust will pollute the environment, too many vehicles will make urban traffic become congested and traffic accidents caused by vehicles and so on. The frequent occurrence of traffic accidents has the most serious impact on human social safety. How to effectively improve the safety of vehicle driving has become the focus of scientific research in various countries, so the concept of intelligent vehicle came into being. In the research of vehicle intelligence, it is the development trend of automobile intelligence to use multi-sensor (lidar, camera, detector) to match with vehicle. Among them, the research and development of machine vision has become the focus of today's research. Binocular vision is an important branch in the field of machine vision, which realizes the perception of machine environment by simulating biological vision. Therefore, the method of binocular stereo vision is used to realize the extraction of lane line, the identification of obstacles, the measurement of target distance and the construction of minimum safe distance model on structured road. The main contents of this paper are as follows: (1) the construction of experimental data acquisition platform. In this paper, a movable horizontal slide track is designed and installed on a laboratory electric vehicle, and a binocular camera is installed on the slide track, where the camera can move on the slide track and adjust the different baseline distance. Finally, the image acquisition can be completed through this platform. (2) the algorithm of lane line and obstacle detection. The experimental images collected by the platform were preprocessed, and then the lane lines were identified and extracted by the improved HOUGH transform. According to the detection results, the ROI region is extracted, and the approximate area of the obstacle is found by edge detection and image entropy calculation. Then the feature points are extracted by the SIFT feature operator to find the exact location of the obstacle. In order to improve the accuracy of target tracking, this paper uses Kalman filter to track the target and verifies the prediction region with NMI features and entropy. (3) the distance measurement algorithm based on binocular vision is studied. The horizontal parallel binocular vision system is used in the ranging process. According to the principle of binocular imaging, the collected image is processed and the visual difference is extracted to complete the calculation of obstacle distance. Through a lot of experiments, the influence of baseline length at different distance on measurement accuracy is found, and the error rate of obstacle ranging is reduced finally. (4) the establishment of vehicle minimum safe distance model. First, the safety warning system and the braking process are analyzed, then the minimum safety distance model is established according to the braking principle. Finally, the model is simulated in MATLAB, and the experimental results show that the model achieves the desired results.
【學位授予單位】:吉林大學
【學位級別】:碩士
【學位授予年份】:2017
【分類號】:U463.6;TP391.41
【相似文獻】
相關期刊論文 前10條
1 祝世平,強錫富;基于坐標測量機的雙目視覺測距誤差分析[J];電子測量與儀器學報;2000年02期
2 汪倩倩;申忠宇;張亮;;基于雙目視覺模型的彩色目標實時檢測與定位[J];南京師范大學學報(工程技術版);2008年01期
3 單寶華;申宇;;基于雙目視覺的拉索模型自由振動監(jiān)測試驗[J];土木工程學報;2012年11期
4 許明杰,王洪榮;雙目視覺中恢復模型點的方法──公垂線法[J];指揮技術學院學報;1997年02期
5 余洪山,王耀南;主動立體雙目視覺平臺的設計與實現(xiàn)[J];工業(yè)儀表與自動化裝置;2004年01期
6 王榮本;張明恒;石德樂;;雙目視覺技術在目標測量中的應用[J];公路交通科技;2007年02期
7 何秋奇;楊玉珍;岳曉峰;孫婷婷;張耀平;;基于雙目視覺的塑料零件數(shù)模重構研究[J];機械工程師;2009年02期
8 李海軍;徐鳳生;張建臣;;基于雙目視覺自動識別系統(tǒng)的應用研究[J];長春理工大學學報(自然科學版);2007年03期
9 張愛華;姚孝國;;雙目視覺脈搏圖像三維信息獲取的研究[J];機械與電子;2009年10期
10 張頌;謝永杰;;基于雙目視覺的空間脫靶量測量技術研究[J];光學與光電技術;2013年04期
相關會議論文 前7條
1 孟環(huán)標;楊興強;李雪梅;;基于雙目視覺的測量系統(tǒng)誤差分析[A];第五屆全國幾何設計與計算學術會議論文集[C];2011年
2 朱宗磊;殷福亮;;基于雙目視覺的目標檢測與跟蹤系統(tǒng)[A];2010年通信理論與信號處理學術年會論文集[C];2010年
3 王榮本;張明恒;毛曉燕;;雙目視覺技術在環(huán)境探測中的應用[A];中國宇航學會深空探測技術專業(yè)委員會第二屆學術會議論文集[C];2005年
4 張靜;李琳;劉曉平;;基于雙目視覺的虛擬手的實現(xiàn)[A];計算機技術與應用進展——全國第17屆計算機科學與技術應用(CACIS)學術會議論文集(上冊)[C];2006年
5 吳濤;劉欣;;基于單雙目視覺信息融合的障礙檢測方法[A];中國宇航學會深空探測技術專業(yè)委員會第二屆學術會議論文集[C];2005年
6 孫明;;基于雙目視覺的移動機器人避障研究[A];中國農業(yè)工程學會2011年學術年會論文集[C];2011年
7 杜英魁;韓建達;唐延東;;基于雙目重建不確定性的動態(tài)特征濾除[A];第七屆全國信息獲取與處理學術會議論文集[C];2009年
相關博士學位論文 前3條
1 許杰;基于雙目視覺和二維熵樹木信息點提取及生長量反演研究[D];東北林業(yè)大學;2015年
2 陳學惠;基于移動機器人雙目視覺的井下三維場景重建方法研究[D];中國礦業(yè)大學(北京);2012年
3 陳希章;基于雙目視覺的弧焊機器人焊縫三維信息獲取研究[D];上海交通大學;2007年
相關碩士學位論文 前10條
1 麥樹彬;基于雙目視覺的色盲輔助技術的研究[D];華南理工大學;2015年
2 徐宏;基于雙目視覺和稀疏表示的智能倒車控制研究[D];西南交通大學;2015年
3 李輝;田間機器人的雙目視覺系統(tǒng)誤差分析研究[D];內蒙古大學;2015年
4 逯沛龍;基于FPGA的雙目視覺圖像預處理系統(tǒng)設計與實現(xiàn)[D];南京理工大學;2015年
5 王婷婷;基于GPU的雙目視覺運動目標檢測跟蹤算法研究[D];哈爾濱工業(yè)大學;2015年
6 王昌盛;基于ORB算法的雙目視覺測量研究[D];哈爾濱工業(yè)大學;2015年
7 趙程;具有圓面特征的非合作目標雙目視覺位姿測量[D];哈爾濱工業(yè)大學;2015年
8 路研研;在軌航天器雙目視覺圖像采集與處理系統(tǒng)研制[D];哈爾濱工業(yè)大學;2015年
9 曹之樂;基于雙目視覺的焦點定位方法研究與應用[D];重慶理工大學;2015年
10 王洪偉;基于雙目視覺的紅綠燈路口安全距離預警研究[D];安徽工程大學;2015年
,本文編號:2416304
本文鏈接:http://sikaile.net/kejilunwen/ruanjiangongchenglunwen/2416304.html