Abstract: In pulse three-dimensional information detection based on binocular vision pulse image sensor, the extraction of pulse image feature points is the critical foundation. In order to improve the location accuracy of the pulse image feature points, a novel algorithm is proposed which can realize pulse image feature points’ sub-pixel location. The algorithm consists of three steps. The first step is image preprocessing, the second step is sub-pixel edge detection based on gray moment, and the last step is to get circular mark central points by the least squares fitting method and complete the pulse image feature points extraction. The experimental results show that the algorithm can accurately locate feature points, with accuracy of central point location up to 0.03~0.04 pixels, and it lays a good foundation for the stereo matching and three-dimensional reconstruction of the pulse image.

Key words: pulse image, edge detection, sub-pixel, least squares fitting method

摘要： 在基于双目视觉脉搏图像传感器的脉搏三维信息检测中，图像特征点的提取是关键基础。为了提高脉搏图像特征点提取的精度，提出一种新的对脉搏图像特征点进行亚像素定位的算法。在完成脉搏图像的预处理后，对脉搏图像进行基于灰度矩的亚像素边缘检测，结合最小二乘拟合法得到圆形标志中心点，最终实现脉搏图像的特征点提取。实验结果表明该算法能够准确定位脉搏图像特征点，精度可达到0.03~0.04像素，为后续双目视觉的立体匹配及脉搏三维形态的重构奠定了基础。

关键词: 脉搏图像, 边缘检测, 亚像素, 最小二乘法拟合