单目结构光测量中参数自适应标定方法

针对结构光测量时手工调节相机参数易导致图像质量的随机性，进而导致标定精度的下降及调参环节重复繁琐的问题，提出了一种结构光测量的相机标定自适应调参方法。首先，设计了一套相机自动调参装置，根据三个调参环调节时的图像变化机理，选取标定板面积占比、图像清晰度及图像对比度三个参数，分别定义焦距调节环、对焦环及光圈调节环。其次，为实现相机的高精度标定自适应参数调节，针对传统Brenner函数的缺陷，通过一种改进的Brenner自动阈值函数，实现对图像清晰度的精确快速对焦，并对标定板图像感兴趣区域的前景与背景进行分割，计算图像对比度，根据标定的重投影误差，确定标定相机的最优调节区间，以自适应调参搜索控制方法进行标定参数调节。最后，为提高自适应标定调参时电机的搜索定位速度，建立了调参函数模型，并通过物距及像素长度自适应计算焦距。三维测量调参实验表明，相比于手工标定，基于提出的调参方法能在5 s内实现自动调参，且重投影误差减少66.57%。

Parameter adaptive calibration method in monocular structured light measurement

Aiming at the problem that manual adjustment of camera parameters in structured light measurement was easy to lead to the randomness of image quality, which leads to the decline of calibration accuracy and the repetition of parameter adjustment, an adaptive parameter adjustment method for camera calibration in structured light measurement was proposed. Firstly, a set of camera automatic parameter adjustment device was designed. According to the image change mechanism of three parameter adjustment rings, three parameters of calibration plate area ratio, image clarity and image contrast were selected to define the focal length adjustment ring, focusing ring and aperture adjustment ring respectively. Secondly, in order to realize the high-precision calibration and adaptive parameter adjustment of the camera, aiming at the defects of the traditional Brenner function, an improved Brenner automatic threshold function was used to realize the accurate and rapid focusing of the image definition, segment the foreground and background of the region of interest of the calibration plate image, calculate the image contrast, and according to the calibrated reprojection error. The optimal adjustment interval of the calibration camera was determined, and the calibration parameters were adjusted by the adaptive parameter adjustment search control method. Finally, in order to improve the search and positioning speed of the motor during adaptive calibration, the parameter adjustment function model was established, and the focal length was calculated adaptively through the object distance and pixel length. The three-dimensional measurement parameter adjustment experiment shows that compared with manual calibration, the proposed parameter adjustment method can realize automatic parameter adjustment within 5 s, and the reprojection error is reduced by 66.57%.