双目标定中棋盘格图像角点样本筛选及标定

摘要

针对摄像机在标定时采集到的图片样本存在样本不理想而导致标定结果不准确的问题．文中基于棋盘标定板的特性提出了一种棋盘格图像样本筛选方法．利用棋盘格中角点的排序的条件，针对图片样本的合法性进行检查，排除角点检测不准确的样本，从而达到提高标定准确度的目的．研究表明：对于同一组样本数据与同样的标定方法，利用筛选后的样本进行标定结果的相对误差减小到0．55％，成功率达到90％以上，提高了样本的可用性与标定的准确度．%Less-than-ideal samples tend to cause inaccurate calibration results.A sample screening method is proposed based on the characteristics of the calibration chessboard. In order to improve calibration accuracy,the samples with inaccurate corner points are removed by using the sort order for corner points in the chessboard and the legality of picture samples.The results show that the probability of successful calibration is higher than 90% and the relative error is reduced to 0.55% using the filtered sample corners with the same set of samples and by the same calibration method.The method improves significantly the availability of samples and the accuracy of calibration.