Abstract: Aiming at obtaining the accurate detection results of design pattern, an optimization method for design pattern identification based on the grammar production is presented for solving the variant problem of design pattern. The method focuses on role attributes of participants and relationships of design pattern instances by visual grammar production, and identifies the overlapping patterns and the additional relations between the participant roles by adding the feature information of grammar production. Experiment results show that the proposed method can reduce the false positive results and the false negative results. F-score index comparison with other well-known algorithms indicates the effectiveness and merits of the proposed method.