Pictures, as auxiliary tools in teaching, are closely associated with teaching knowledge in teaching practice. Picture resource library is a collection of picture resources and a key channel to link different teaching resources. Relying on the "Smart System of International Chinese Education", this paper generates teaching pictures applicable to the field of international Chinese language education by using a multimodal large model, and determines a picture annotation system based on the idea of "Precise Annotation and Interconnection" in order to form a picture resource system, which constructs a hierarchical system of language and picture elements. A hierarchical network between language elements and picture elements is constructed. The resource library contains picture resources that are highly related to different language units, and on the basis of the existing pictures, the picture labeling system is formed with reference to the typical teaching methods in the international Chinese language education field, so as to enhance the application effect of picture resources in teaching. This resource processing process provides an example for picture generation and labeling based on multimodal large models, and provides a demonstration application for the construction of generative resource libraries.