Aimed at solving the problem about obtaining pose information and calculating the theoretical center of solder joints for BGA images with missing solder joints in the task of BGA chip visual inspection，an improved point pattern matching algorithm was proposed after analyzing the deficiency of the existing point pattern matching algorithm. The algorithm establishes a solder joint distribution matrix model base on the arrangement of BGA solder joints, and uses the metric space to expand the grid, maps it to the image space through affine transformation, uses the grid as a clue to determine the matching relationship, and finally uses the point set registration to complete accurate calculation，get the pose information and the theoretical center of the solder joint. Through simulation experiments, it is proved that the method can adapt to the situation of missing solder joints. The position error is 1.6% and the accuracy is improved by 26.4% compared with the traditional method.