Abstract: Focusing on dense matching in photogrammetry, an image matching method by combing the triangle network constraint with pyramid strategy has been proposed in this paper.Firstly, feature points are extracted in each layer, and then image matching is done under the constraint of a Delaunay triangle network, which has been constructed by the matching result of a higher layer.The pyramid strategy itself shows a coarse-to-fine process and meanwhile Delaunay triangle network can pass a powerful match control from the higher layer to the lower layer.Experiments shows that the proposed method can get a reliable, high-accuracy and dense match result and the generated point cloud can describe the topography well.