基于相位编码技术的双目视觉亚像素匹配算法的研究

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,用所述光栅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,并拟合成连续曲线,从而实现了两幅图像相同相位曲线的匹配,再根据极约束条件,得到两幅图像点,采用插值算法计算出点的精确匹配。通过对实际物品的测试实验和标准量具的测试实验,实验验证了该方法进行立体匹配的高效性和准确性。

The Research of Algorithm Based on Phase Encoding of Binocular Vision Sub-Pixel Matching

In the three-dimensional vision, compared with several common three-dimensional matching, and the existing problems, several methods are proposed in one of the high-dimensional financial matching. First, the computer constructed grating which is used to solve phase projects onto the object by the projector, then the computer constructed a series of black and white encoding grating are projected by the projector onto the object; then distorted grating images collected by the camera computer, using solution phase by phase grating solution fold in ？ -ID, ID] interval principal value of phase with grating encoding method described by the phase of the cycle phase started, the two added together are the true phase value, and fitted into a continuous curve, thus achieving the same phase curve of the two images match, then under very constrained conditions, points in two images obtained using the interpolation algorithm to calculate the exact match point. Through the actual test items and standards for measuring the test experiment experiment to verify the method of stereo matching efficiency and accuracy.