基于轮廓特征理解的城市道路图像深度估计介绍.doc

  本文关键词：基于轮廓特征理解的城市道路图像深度估计，由笔耕文化传播整理发布。

基于轮廓特征理解的城市道路图像深度估计 丁伟利* 李勇 王文锋 曹秀艳 燕山大学电气工程学院 河北省秦皇岛市 066004 摘要实验结果本文提出的算法能够有效地内部的渐变深度信息。 关键词?? Depth Estimation of Urban Road Image Based on Contour Understanding Ding Weili, Li Yong, Wang Wenfeng Cao Xiuyan Institute of Electrical Engineering, Yanshan University, Qinhuangdao ,Hebei,066004, China Abstract Estimating the depth of the road scene is a key step for intelligent transportation and robot navigation. This paper presents a depth estimation algorithm for single static road image. The algorithm obtains a series of closed areas of the image bede on the edge growing segmentation algorithm firstly; then we compute the statistic characteristics of each region, including color, area, position, straight lines, vertical lines, and parallel lines. Based on these characteristics, the road vanishing point was further estimated, and the sky, road and vertical regions are estimated. Finally, the depth of the road image is estimated by using vanishing point and the statistic characteristics. Experimental results show that the proposed algorithm can effectively estimate the road vanishing point, as well as the road depth information. Keywords Computer vision; Depth estimation; Image understanding; 3D space estimation; Single image OCIS Codes XXX.XXXX; XXX.XXXX; XXX.XXXX（需给出具体的栏目分类，而不是总的栏目号，如320.320无效） 1引言 从单幅图像恢复场景深度信息和三维结构是计算机视觉一项重要的研究内容，它通过提取与深度相关的视觉线索和一些先验知识完成深度估计，有助于实现基于单像机的场景理解、目标定位、三维测量和重建，在机器人导航、智能交通和增强现实场景注册领域有广泛的应用前景[1-2]。 现有的基于单幅图像的深度估计算法主要利用三维空间的

  本文关键词：基于轮廓特征理解的城市道路图像深度估计，，由笔耕文化传播整理发布。