摘要
为实现仅用一幅二维图像,在系统参数未知且可变条件下,在真正的三维欧氏空间重构三维场景,其解决的首要关键问题是研究合适的编码结构光照明方法.文章采用伪随机彩色空间编码的可靠模型算法设计该主动视觉系统的多种编码投影模板,探索实现三维场景表面高分辨率伪随机彩色空间编码的特征化新技术,该编码窗口唯一性能唯一辨识待重构场景表面上每一特征点,解决场景重构特征点的匹配瓶颈难题,其不标定三维欧氏重构结果证实该文设计方法具有有效性.
In order to reconstruct the real euclidean 3D scene only with single image captured by uncalibrated camera whose parameters are unknown and adaptive to the varying scene,the principal point is to research appropriate encoded structure light. In this active vision system, this paper designed several encoded templates adopting credible algorithm of pseudo-random color encoded array to actualize new high ratio characteristic technique of the surface of 3 D scene. The single trait of codes window could distinguish each feature point of 3D scene to be recovered, which resolved the bottleneck problem of feature matching. The uncalibrated result of the euclidean reconstruction of3D scene comfirmed the reliability of the present method.
出处
《安徽大学学报（自然科学版）》
CAS
北大核心
2011年第1期69-73,共5页
Journal of Anhui University(Natural Science Edition)
基金
国家自然科学基金资助项目(50275049
60931002)
安徽大学青年科学研究基金资助项目(2009QN030B
XJ200925)
关键词
单幅图像
不标定三维重构
伪随机编码
彩色编码模板
特征点匹配
single image
uncalibrated 3D reconstruction
pseudo - random array
multicolor encoded template
feature matching
