基于 Hessian 矩阵的多结构光条纹中心快速提取方法

摘要

Fleetly and accurately extracting the center line of light stripes of image is needed when measuring the dimension of parts. Usually, multi-structured light stripes are used in dimension measurement system. In order to rapidly extract the sub-pixel center and realize the identification of multi-structured light stripes, a method based on Hessian matrix is proposed. Firstly, the structured light stripes are distinguish and coded. Secondly, the extreme points on every of the structured light stripes is calculated. Finally, the sub-pixel center of the structured light stripes is determined according to the extreme point. Experimental results show that this method can reduces the calculative quantity and can improve the computational effectively.%使用结构光方法对物体进行三维检测时，需要对图像中结构光条纹中心快速、准确提取。为提高检测速度，可向待测物体同时投射多结构光条。为实现多结构光条纹识别和亚像素中心快速提取，本文提出一种基于 Hessian 的多结构光条纹亚像素中心提取方法。该方法首先对多结构光条纹进行识别，根据识别的结构光条纹计算各条纹上的极值点，最后在极值点像素位置附近区域确定结构光条纹的亚像素中心。实验表明，该方法减小了 Hessian 矩阵算法的运算量，提高了结构光亚像素中心提取速度，能满足多结构光条纹识别和亚像素中心快速提取的需要。