Abstract: To enhance the accuracy of design pattern detection, an automated detection method that incorporates microstructures is proposed. Firstly, grounded in the characteristic description of constraint satisfaction problems, this method categorizes and summarizes the microstructures of design patterns. Secondly, definitions of design pattern features and microstructures are established. A phased preprocessing methodology for candidate design pattern instances is introduced. Finally, feature vectors are constructed, and subsequently, machine learning techniques are applied to classify the design patterns. Furthermore, two-phase experiments are designed to detect both standard design patterns and their variants. The experimental results validate the effectiveness of the research, demonstrating significant improvements in design pattern detection accuracy.