[Purpose/significance] Under the background of increasingly rich multimodal information resources, cross modal information organization has important value for the sequential storage and sharing of intangible cultural heritage information resources. This paper proposes a method of constructing intangible cultural heritage cross modal knowl⁃edge graph for image and text data. [Method/process] This paper combined multimodal information extraction and knowledge graph technology to reorganize the knowledge of intangible cultural heritage digital resources. Firstly, the schema layer of intangible cultural heritage cross modal knowledge graph is designed from the content and modal di⁃mensions of intangible cultural heritage digital resources to provide top-level design for graphic data organization. Sec⁃ondly, this paper discusses the knowledge extraction scheme of image and text resources in the field of intangible cultur⁃al heritage, obtains the intangible cultural heritage knowledge in the mode of image and text, and verifies the effective⁃ness of the knowledge extraction method. Finally, a series of cross modal knowledge linking methods for different scenes are used to extract the cross-modal relationships between intangible cultural heritage graphic entities. [Result/
conclusion] Taking the intangible cultural heritage project of Chu opera as an example, the cross-modal knowledge graph of Chu opera was constructed through the combination of multiple data sources and technologies, which verified the feasibility of the construction method proposed in this study, and had a certain reference value for the digital con⁃struction of intangible cultural heritage.