An area of knowledge extraction is mobile user data mining. It is concerned with methods and algorithms on extracting interesting knowledge from mobile users through the data they have generated. These data are such as their user movement database and communication history. In group pattern mining, group patterns from a given user movement database is found based on spatio-temporal distances. Static objects are such as walls are present in the mobile environment. In this paper, we propose a method of group pattern mining through a user movement database with static objects defined to ensure the accuracy of result when static object exist. Our performance evaluation witnessed a reduction of group pattern found after static objects are defined in user movement databases compared to without. It proves that mobile users that are separated by static object can be detected and prevented from returning them as a valid group pattern.