为解决实际测量中求光条中心速度慢的问题,提出一种基于阈值法和Hessian矩阵的改进算法.首先,使用均值滤波以削弱图像噪声,并通过设定阈值来找出光条中心粗略位置,而后利用高斯函数的可分离性和对称性获得光条中心点的Hessian矩阵,以此求出定每一个粗略中心点的法线方向,最后,经过光条中心点在其法线方向上的泰勒级数二次展开得到亚像素坐标点.实验表明该算法能有效降低噪声对图像的影响,极大提高Hessian矩阵的求取速度,同时精度也可达到亚像素级.%An improved algorithm that based on threshold and Hessian matrix is proposed to solve the problem of extracting speed of light stripe center.Firstly,the image noise was reduced by mean filtering and extract structured light stripe center roughly by threshold method.Then obtain Hessian matrix by using the separability and symmetry of Gaussian template and the normal directions for the points of light stripe can be determined.Finally,the sub pixel coordinates of the center of the light strip are acquired by Taylor series expansion in the normal direction.Experimental results indicate that this algorithm can extract center of structured light stripe accurately and the separability and symmetry of Gaussian template can reduces the computing task drastically and implements high-accurate center extraction of structured light stripe with high speed.
展开▼
