Abstract: A matching algorithm of coronary artery angiogram based on anatomic model and epipolar line constraint is proposed. Through this the feature points can be mapped and matched. The feature points are extracted from vascular skeleton by matching the anatomic structure to set up the geometrical transfer matrix. Other key points of vascular skeleton are matched by epipolar line constraints. The matched feature points are further optimized by correlative algorithm to improve the veracity of the matched points. Meanwhile, the width of the vessel obtained by the edge-detected method is helpful to reconstruct the three dimensional coronary arterial tree. The processing algorithm provides reliable diagnosis information.