区别于图像的简单几何特征，利用图像的全局特征描述子──图像矩特征作为图像特征信息，推导了矩特征变化量与相对位姿变化量之间的关系矩阵，即图像雅可比矩阵．针对平面视觉伺服控制，利用了所推导的基于矩特征的图像雅可比矩阵，在不需要知道目标的成像高度、摄像机焦距及不需要精确标定的摄像机外部参数的情况下实现了基于图像的视觉伺服控制．最后给出了仿真与实验例子，表明了矩特征的有效性． Different from the simple geometric features of the image, the global mappings of image features as the image feature information are used to deduce the relation matrix between the variation of the moment feature and the relative pose variation, that is, the image Jacobian matrix. For planar vision servo control, the derived Jacobian matrix of images based on moment features is used to realize image-based visual servoing without knowing the target’s imaging height, camera focal length and camera external parameters without precise calibration control. Finally, the simulation and experimental examples are given to show the validity of the moment features.