对逐点图像畸变校正算法进行分析和研究。分析了基于四点共线交比不变性的逐点畸变校正算法的校正精度，指出该算法由于初始使用图像点含有误差导致计算其他图像点产生了精度改变，同时给出了计算结果误差较小时的图像点位置和相对距离的选取方法。基于上述分析，提出了基于消隐点共线约束的逐点畸变校正算法。该算法利用消隐点来提高直线拟合精度，利用共线特征来约束校正图像点精度。该算法不仅可以优化基于四点共线交比不变性算法校正的图像点，同时也可以优化初始使用的图像点，从而提高所有图像点畸变校正精度。基于MATLAB的仿真实验显示: 对于400万像素的镜头，校正后图像点最大误差是初始图像点噪声的29.05倍。针对实物图像对基于四点共线交比不变性法校正的图像点，基于消隐点共线约束法校正的图像点和未校正采集图像点的交比值进行对比，结果表明本文提出算法优于四点共线交比不变性算法的结果。

Point wise correction algorithms are researched and analyzed. The correction accuracy of a point wise distortion correction algorithm based on four-point collinear cross ratio invariability is investigated. It points out that the initial used image points which contain errors will cause the accuracy change in the calculation of other image points. Then, it gives the selection method for image point locations and relative distances when the calculation error is smaller. Based on the above, this paper proposes a point wise distortion correction algorithm based on vanishing point collinear constraint. The algorithm uses vanishing points to improve the linear fit accuracy and the collinear feature to constrain the corrected distortion accuracy. It not only optimizes the image point errors of the distortion correction algorithm based on a four-point collinear cross ratio invariance, but also the initial image point errors, so that the accuracy of distortion correction of all image points is improved ultimately. The simulation based on MATLAB shows that the maximum error of corrected image points is 29.05 times that of the initial image point noise for four million pixels. In contrast to the physical image using the cross ratios from the correction algorithms based on four-point collinear invariant cross ratio and vanishing point collinear constraint as well the uncorrected image, it indicates that the performance of proposed algorithm is better than that of the algorithm based on four-point collinear cross ratio invariance.