基于视觉的内窥镜端部位姿模型分析

Vision-based orientation model analysis of handling part of endoscope

针对在假设内窥镜端部固定不变的条件下进行基于光纤光栅传感阵列的内窥镜柔性杆的三维重建存在的问题,提出了基于标志点的内窥镜端部位姿视觉检测模型;利用视觉技术和通过标志点的空间位置信息来获取内窥镜端部位姿的方法,建立了基于主元分析方法及最小二乘法分析标志点位置信息误差对内窥镜端部姿态的影响的方法。模拟分析结果表明,当三个标志点的位置信息误差满足高斯分布且方差在0～3mm 之间时,其姿态的最大误差角度为0.59°。通过实验证实了理论计算方法能够有效地利用带有误差的位置信息求解内窥镜端部位姿及误差的大小,为微创手术器具的位姿检测提供了理论和实验基础。

Aiming at the problems caused when reconstructing the shape of an endoscope＇ s flexible rod using a fiber bragg grating （FBG） sensor array on the assumption that the handling part of the endoscope is fixed, the paper proposes an orientation detecting model based on binocular vision, and establishes a method for mathematical analysis of the effects of markers＇ position information errors on the orientation of the handling part of the endoscope with the principal component analysis （PCA） and the least square method （LSM）. The simulation experiment demonstrates the maximum angle error of orientation is about 0.59 degree when the position errors satisfy the Gaussian distribution with the mean being zero and the standard deviation being 0 - 3mm. The real experimental results verify this method can robustly solve the orientation of an endoscope using the position information of markers with position errors, and this provides a theoretical and experimental basis for orientation measurement of instruments for invasive surgery.