The invention discloses a uniform GMM image matching method based on fusion space and feature information. A Gaussian mixture model which effectively fuses spatial information and feature informationis adopted. The Gaussian component weight coefficient is not added with the priori information, the Gaussian component contains the spatial information and the feature information at the same time, and the solution of the matching relation is determined by the spatial information and the feature information at the same time. When the model points in the model point set and a certain data point inthe data set meet the condition that the feature descriptors are similar or the spatial positions are aligned, the two points are likely to form a pair of matching, so that the image matching accuracyis improved.