以關聯式動態規劃法做雙眼立體視覺偵測; Binocular stereo vision detection based on the associated dynamic programming

NCU Institutional Repository > 資訊電機學院 > 資訊工程研究所 > 博碩士論文 > Item 987654321/9804

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/9804

题名:	以關聯式動態規劃法做雙眼立體視覺偵測;Binocular stereo vision detection based on the associated dynamic programming
作者:	林育正;Yu-Cheng Lin
贡献者:	資訊工程研究所
关键词:	動態規劃;雙眼立體視覺;binocular stereo vision;dynamic programming
日期:	2009-07-02
上传时间:	2009-09-22 11:56:36 (UTC+8)
出版者:	國立中央大學圖書館
摘要:	隨著經濟的成長，機動車輛愈來愈多，因而交通事故也愈來愈多。有鑑於此，發展車輛輔助安全駕駛的議題，也就愈顯示其重要性。市區的行人防撞是其中一項重要議題。在本研究中，我們提出一個利用雙眼立體視覺來取得深度資訊的方法，並應用於車輛前方的行人偵測上，以避免己車撞及行人。另外在倒車輔助的應用上，我們使用平面投影轉換的立體視覺方法來偵測障礙物，以避免倒車時碰撞到障礙物。在車輛前方的行人偵測中，我們先根據兩張影像平均亮度的比值來調整兩張影像的亮度，使其較為相近。再根據水平投影累積資訊調整兩張影像的垂直位置，使對應的掃描線能夠儘量接近極線 (epipolar line) 的位置。之後再透過我們所提出的關聯式動態規劃法計算出像差圖；再透過形態學平滑化，將雜訊消除。接著產生連結區塊，將行人或障礙物框選出來，並根據像差值，求得該物體與攝影機之距離。在倒車輔助方面，我們先利用相機校正取得兩相機各相對於地面之轉換矩陣，再將左影像透過反轉換轉至右影像平面。之後將兩影像相減平方，並利用形態學平滑化去除雜訊。最後產生連結區塊，框出障礙物並利用垂直邊資訊，框出較精確的地面上障礙物範圍。 In these few decades, the vehicle number is rapidly increasing due to people’s incomes increasing. In addition to the vehicle number, more factors of road situation, driving environment, and human attention result in a large amount of traffic accidents and casualties. If there is a mechanism to help the driver to detect the road situation and driving environment, and then provide some useful information to the driver in these situations, the danger is therefore avoided. It is important to develop real-time automotive driver assistance systems. Pedestrian collision avoidance is one of the important issues. In this study, we propose a method to get depth information with binocular stereo vision, and apply to pedestrian detection in front of the vehicle. Moreover, we use homography to detect obstacles to avoid the close collision. In pedestrian detection system, we first adjust the illumination and vertical position of the image pair. Then, we use associated dynamic programming to generate disparity map. Thirdly, we use morphology to reduce noise. Finally, we generate connected component to detect pedestrians or obstacles and estimate distances based on the disparity. In parking assistance system, we first use camera calibration to get the transform matrices between the cameras and the ground coordinate system. Then, we transform left image into the right image plane via the ground coordinate system. Thirdly, we subtract the right original image and the re-projected image to generate a difference image. Fourthly, we use morphology to reduce the noise in the difference image. Finally, we generate connected component from the difference image to detect obstacles.
显示于类别:	[資訊工程研究所] 博碩士論文

文件中的档案:

档案	大小	格式	浏览次数

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....