Abstract: Existing few-shot Chinese character generation methods struggle to effectively decouple font style and content using a limited number of reference characters, resulting in subpar font generation. To address this issue, a novel few-shot Chinese character generation method based on contrastive learning and square-block transformations for data augmentation is proposed. By leveraging the concept of contrastive learning, we perform style contrast and content contrast on the output images, achieving better adaptive decoupling of font style and content. To further enhance the performance of few-shot font generation, a data augmentation method based on square-block transformations is introduced. Through geometric transformations of the square-block regions, we embed the global structural information of the font, improving the ability to decouple font style and content. The effectiveness of the proposed method has been validated through extensive experiments. Experimental results show that compared to the current state-of-the-art few-shot font generation methods, the proposed method demonstrates significant advantages in both performance metrics and the quality of generated fonts. Notably, when integrated into existing few-shot font generation models and trained with a smaller dataset, the proposed method can significantly improve the performance of existing few-shot Chinese character generation models.