机器视觉中基于彩色伪随机编码特征点匹配的研究

The correspondence between the surface to be recovered and its images based on multicolor pseudo-random array in machine vision

为研究基于单幅二维图像不标定欧氏重构三维场景的理论,介绍一种采用伪随机编码结构光照明主动视觉技术。利用伪随机序列的窗口特性,使被编码结构光照明的场景表面每一个特征点都具有唯一的代码,可以唯一地被辨识。实验表明,该编码结构光主动视觉系统,可以比较容易地解决被动视觉系统中难以解决的特征点匹配问题,实验效果良好。

In order to study the theory of uncalibrated 3D Euclidean reconstruction based on single image. Introduces an effective active-machine technique with structured light illumination which applies pseudo-random color code. Each interest point on scene surface can be identified exclusively according to the window property of pseudo-random array. It is found out that this active machine system can solve the correspondence between images in the passive machine system. The results of the experiment are satisfying.