通过模拟实际光学成像过程以及CCD采样过程,生成了标准图像,实现对亚像素定位算法的客观评价,并提出了一种新的基于灰度面积插值的亚像素定位方法。经实验验证,该方法对直线边缘的平均定位精度达到0.11 pixel,对圆边缘点的平均定位精度可达0.24 pixel,优于梯度质心法和灰度矩法。

The paper proposes a method to build standard images by simulating the process of optical imaging and CCD sampling,which provides an objective standard for the evaluation of sub-pixel location operators. Also a novel sub-pixel location operator based on gray area interpolation is proposed,experiments prove that the average accuracy of the arithmetic is about 0.11 pixel for linear edge,and 0.24 pixel for circle edge,which is better than the operators based on gradient centroid and gray moment.