基于面结构光的双目立体匹配算法研究

Binocular Stereo Matching Algorithm Based on Surface Structured Light

提出一种基于面结构光的四步相移三频外差的双目立体匹配方法。将正弦光栅投射到被测物体上,将左右目CCD采集的光栅图像进行四步相移三频外差算法相位展开,然后在极线准则的条件下,确定左右视角主值相位相等的点完成双目的立体匹配。实验证明,该方法具有实时性强、匹配精度高,在摄像机精度较低的情况下,具有较好的匹配效果。

This paper proposes a surface structured light and the binocular stereo matching algorithm based on four-step phase-shifting and tri-band heterodyne algorithm. First, put the sinusoidal gratings onto object. And than, get the image of the CCD on both side and unwrap the phase of the image by using the four-step phase-shifting and tri-band heterodyne algorithm. Finally, complete the binocular stereo matching by finding the equal points in the left and right unfolded phase image. Experiments show that this method has strong real-time, high matching accuracy and has a higher matching accuracy even if the CCD accuracy is low.