【作者】 王立娜；

【导师】 朴燕；

【作者基本信息】 长春理工大学 ， 电子科学与技术， 2010， 硕士

【摘要】 从全景图像中取深度信息,提取过程简单,而且无须进行相机标定,正成为国内外三维图像领域研究的热点。比较完整的深度提取过程包括视图的提取、图像预处理、立体匹配、视差图获取和深度确定等几个部分。本文以从全景图像中提取深度信息为中心。首先,对全景图像技术及其发展进行了详细的阐述,分析并推导出从全景图像中提取深度信息的计算方法；然后重点对全景图像“视图“对之间的视差进行深入的研究。在立体匹配方面,针对全景图像技术的成像特点,采用了基于角点特征的双向匹配算法,并加入Marr立体匹配约束条件,以获得更多的最佳匹配点,然后对得到的视图对间的视差进行分析。最后,利用深度和视差的关系,求出目标点的深度信息。更多还原

【Abstract】 Taken depth information from Integral Imaging, the extraction process is simple and doesn’t require camera calibration, and three-dimensional image reconstruction is becoming a hot area of research at home and abroad.A complete process of depth extraction includes the following parts:extraction of view image, image pre-processing, stereo matching, disparity map and depth extraction. This paper carried out the research centering at depth extraction from Integral Imaging.Firstly,we described the Integral Imaging technique and its development in detail; Analyzed and derived the calculation method of depth information extraction from Integral Image; Then focused on the research of disparity analysis between the view image pairs. In stereo matching, by analyzing imaging features of integral imaging, use corner-based feature two-way matching algorithm, and add Marr stereo matching constraints to obtain more the best match points, and then conduct a parallax analysis between view images. Finally, by the relationship between depth and parallax, it finds out depth information of target points.更多还原