The determination of conjugate points in a stereo image pair, i.e. image matching, is the critical step to realize automatic
surveying and recognition in digital photogrammetric processing. The accuracy of image matching is closely related to
specific matching algorithm as well as images. In this paper, the qualitative and quantitative relationships between the
matching accuracy and the image metrics are studied at the basic of Least Squares Image Matching algorithm (LSIMA).
Firstly, the algorithm is deduced mathematically, and then the main image metrics affecting the matching accuracy are
presented, including total variation (TV) metric and difference of signal-to-noise ratio (DSNR) metric. Subsequently,
variations of matching accuracy with TV and DSNR are analyzed, and mathematical model between them is developed.
Studies show that the matching accuracy presents the natural exponential rule along with TV and DSNR of image pairs.
Besides, parameters of the model are estimated and the model is verified by simulation experiments. Finally, the
correctness of the model is verified using real remote sensing images. Experimental results demonstrate the robustness
and accuracy of the proposed model.