Abstract: Point set matching is one of the classical NP problems in computer vision and pattern recognition. Membrane computing is an emergent branch of natural computing, which aims to abstract innovative computing models or computing ideas from the structure and function of a single cell or from complexes of cells, such as tissues and organs. On the basis of membrane optimization algorithms with hierarchical structure and the feature of the point set matching problem, a novel point set matching algorithm was proposed. In this algorithm, three new heuristic search rules were introduced, by which matching rate increased to some extent. Compared to the traditional optimization algorithms, the algorithm exhibited a better global search capability, thus a better solution for point set matching problem was obtained. Experimental results illustrate that the proposed algorithm is effective on both matching rate and stability.