一种基于人工地标解码的定位方法

A Location Method Based on Artificial Landmark Decoding

针对当前机器人摄像头中心位置无法快速定位问题,提出人工视觉地标的编码以及解码定位方法。首先进行编码片以及人工视觉地标编码方式设计,采用回溯生成算法从单个编码片编码方式延伸到多个编码片拼接,利用M阵列思想实现全局唯一人工视觉地标矩阵;然后进行人工视觉地标解码,针对编码片利用轮廓链进行编码组件提取,利用HSV颜色空间进行颜色提取从而实现码字识别。对人工视觉地标中心进行定位识别,通过编码块中心线拟合得到的特征点进行二次曲面拟合消除畸变,得到人工视觉地标下摄像头中心所在编码片的编码形式、码字以及摄像头中心世界坐标。实验证明,插值法得出的像素点坐标在X方向上最大误差为3个像素,最小误差为0个像素,Y方向上最大误差为2个像素,最小误差为0个像素,对比传统棋盘格标定定位提高了0.123mm/像素精度。

Aiming at the problem that the center position of the current robot camera cannot be quickly located,a method of encoding and decoding positioning of artificial visual landmarks is proposed.First,design the coding method of the code slice and the design of the artificial visual landmark coding method;use the retrospective generation algorithm to extend from a single coding slice to the splicing of multiple code slices;use the idea of M array to realize the globally unique artificial visual landmark matrix.Secondly,the decoding of artificial visual landmarks is carried out.For the coded slices,the contour chain is used to extract the coding components;the HSV color space is used for color extraction to realize code word recognition;the artificial visual landmark center is positioned and recognized;the center line of the coding block is fitted The obtained feature points are fitted to the quadric surface to eliminate the distortion to the final pixel coordinates.Obtain the encoding form,code word and world coordinates of the camera center of the camera under the artificial visual landmark.The pixel coordinates obtained by interpolation have a maximum error of 3 pixels in the X direction and a minimum error of 0 pixels in the Y direction.The maximum error is 2 pixels,and the minimum error is 0 pixels.Compared with the traditional checkerboard calibration positioning,the accuracy is improved by 0.123mm/pixel.