In recent years, named entity recognition (NER) has attracted significant attention in various fields, especially in the clinical medical field, because NER is essential for useful mining knowledge in the clinical medical area. However, there are still some problems in Chinese named entity recognition, such as the complexity of medical texts, word segmentation errors, and incomplete extraction of semantic information. In this paper, we propose a Chinese NER method based on the multi-granularity semantic dictionary and multimodal tree method, which involves the following steps. First, we extract different semantic words using multimodal trees. Next, we extract the boundary information, and finally, perform the multi-granularity feature fusion. Furthermore, we combine the above methods to complete the entity recognition task. From the results of our experimental verification, our proposed model outperforms the current state-of-the-art results.