基于结构光和极线约束的三维重建

A method of 3D reconstruction based on structured light and epipolar constraint

结构光可以简单、快速并且精确地对被测物体进行三维信息提取,从而得到三维数据点云。而在结构光三维信息获取系统中,条纹的准确检测是影响系统精度的关键因素之一。为了改善这一问题,先利用格雷码对图像条纹进行粗定位,再结合相位码准确地对图像条纹进行细定位并利用极线约束把图像匹配过程中的二维搜索降为一维搜索。结构光条纹和极线的交点就是图像的匹配点,对匹配点空间前方交会即可得到物体的三维点云。反复试验表明,该方法能取得较好的效果。

The structured light can be used in extracting the 3D information for obtaining the three-dimensional points from object to be measured, and this method is simple, fast and accurate. However, the key factor of this process is how to inspect the stripes very accurately. To improve the accuracy, this paper firstly locates the stripes of the structured light roughly by use of the Gray code images, and then locates the stripes of the structured light accurately by utilizing the shift code images. Finally turns the two dimensions search into one dimension search in the images match by imposing the epipolar constraint, so the cross points of the stripes of structured light and the epipolars are the match points, finally do space intersection to the match points and obtain the three-dimensional points. We obtain the better effect by the method, which is proved by experiments repeatedly.