基于编码光栅和外极线约束的曲面匹配算法研究

Research of curved surface match algorithm based on coding raster and epipolar constraint

提出了一种新的基于编码光栅和外极线约束的三维曲面匹配算法,该算法有效解决了立体视觉中的图像匹配问题,即左右图像光栅条纹的对应关系确定问题。另外,采用外极线约束将可能的候选点限制在直线分布,即将查找范围由二维降为一维。这种算法极大地减小了图像匹配的运算量,同时减小了错误匹配的概率。实验表明,采用编码光栅投影与外极线约束的视觉测量能够准确有效地匹配出自由曲面的三维立体轮廓。

This paper presented a new curved surface match algorithm based on coding raster and epipolar constraint. With this algorithm, match problem, that is to say, coincidence relation of raster stripe was easily and precisely settled. In addition, epipolar constraint was applied to limit candidate-matching point to a line and search range was minished from two dimension to one dimension. This method saved computational time greatly and miscorrespondences were reduced at the same time. Using coding raster projection and epipolar constraint, free-form 3-D profiles can be matched efficiently and accurately.