在高精度的摄影测量中，图像处理精度对整体测量精度起着至关重要的作用，但成像过程中摄像机对特征点图像的离散化采样，会造成图像与原始信号的失真，从而带来图像处理环节的误差。通过对图像处理过程中特征点总能量、能量分布弥散半径和图像处理窗口的特性分析，并以特征点中心位置、提取误差大小和能量密度函数的标准差之间关系为基础，提出一种针对离散化采样的误差补偿法。该方法仅需标定一次补偿参数，适用于所有摄像机和算法，可显著提升原有图像处理精度。实验证明，对于质心法和高斯拟合法，该补偿法可将图像提取精度提高到0.03 pixel。

In the photographic measurement with high precision, the image processing precision plays a crucial role in the overall measurement accuracy. In the process of imaging, the discrete sampling of the feature points by cameras causes a distortion between the received signal and the original image, which brings errors in image processing. By analyzing the total energy, the dispersion radius of energy distribution of the feature points and the image processing windows, and based on the relation of the feature point center position, extraction error and the standard deviation of the energy density function, an error compensation method is proposed for discrete sampling. According to the method, only once calibration of the compensation parameters is needed for any camera and algorithm, and the image processing precision can be improved significantly. The experiments show that for the centroid method and the Gaussian fitting method, the image processing precision can be increased to 0.03 pixel.