对已有的圆形标志中心定位方法进行改进,提出一种新的基于非线性最小二乘拟合的标志定位方法。并推导出定位算子的不确定性公式,用蒙特卡洛积分方法计算在不同置信水平下的置信区间。实验表明,在摄影比例尺为1:75的条件下,新的定位算子比子像素定位算子有更低的不确定性,精度提高了将近10倍,可以达到0.02个像素左右。
A new center positioning algorithm of circular mark was put forward, which can fit the ellipse parameters by nonlinear least squares. It improved the ellipse edge detection method in the sub-pixel positioning algorithm. Moreover the estimating formula for mark location uncertainty was derived, and confidence interval in a certain confidence level was calculated by use of the Monte Carlo integration. An example certificates that the new lo- cation algorithm can achieve about 0.02 pixel positioning accuracy which rises nearly 10 times than that with the sub - pixel positioning operator when the photogrammetric scale is 1:75.