讨论一个摄影测量网络的三维自动重建。其中，重点讨论了无需相似性量测的多片影像匹配。该匹配算法包括了所有影像作为参考影像的连续考虑和整个网络同时匹配。本文中，三维坐标矢量只用于剔除多余候选点，对精度提高无任何贡献。本方法的精度是通过子像素的特征点提取和多片交会得到的。 Discuss the 3D automatic reconstruction of a photogrammetric network. Among them, the focus is on multi-chip image matching without similarity measurement. The matching algorithm includes the continuous consideration of all the images as the reference image and the entire network to match at the same time. In this paper, the three-dimensional coordinate vector is only used to eliminate the extra candidate points, without any contribution to improving the accuracy. The accuracy of the method is obtained through the sub-pixel feature extraction and multi-intersection.