双目CCD结构光三维测量系统中的立体匹配

在立体视觉中,比较了几种常见立体匹配方法,并针对存在的问题,提出了多种方法融于一体的高效立体匹配方法.首先将计算机构造的用于解相的光栅由投影仪投射到物体上,再将计算机构造的一系列黑白相间的编码光栅也由投影仪投射到物体上;然后将畸变栅线图像由摄像机采集到计算机中,用解相光栅进行解相得到折叠在[-π,π]区间内的相主值,用所述结构光编码方法进行相展开得到相位的周期,二者相加得到真实相位值.利用插值算法将具有相同相位的点拟合成连续曲线,从而实现了两幅CCD图像相同相位曲线的匹配,再根据外极约束条件,得到两幅图像点的匹配.在反向工程中,用该方法对电话接线筒进行三维测量,实验验证了该方法进行立体匹配的高效性和准确性.

Stereo matching in 3D measurement system using double CCD structured light

The establishment of correspondence between two images is the precondition for 3D reconstruction in computer vision. By comparing several commonly used stereo matching methods, a highly effective stereo matching method, which is a combination of several matching methods, is proposed as a solution to the existing problem as detailed below: a series of gratings constructed by computer for phase demodulation is projected onto the object, and a series of black and white coded gratings constructed by computer is then projected onto the object, the distorted grating images are collected into the computer by CCD, the primary value of phase folded in zone is obtained through phase demodulation using the phase demodulation gratings, the period of phase is obtained through phase demodulation using structured light coding proposed, the true phase value is obtained by adding the two, the points with the same phase are fitted into a continuous cure by interpolation to realize the matching between two CCD images with the same curves, and the matching between the corresponding points in two images is obtained according to the epipolar constraint conditions. Experimental results proved the correctness and effectiveness of the method proposed.