Oracle Bone Inscriptions (OBI), the earliest mature Chinese writing system, remain partially undeciphered due to significant glyph evolution across historical periods. Previous AI-driven approaches have primarily focused on directly establishing relationships between OBI and modern Chinese characters, a methodology that inherently suffers from fundamental limitations due to the vast morphological gap accumulated over millennia. To bridge this evolutional divide, we propose a cross-font image retrieval network (CFIRN) that introduces historical font intermediaries (Bronze Inscription, Bamboo Slip Inscription and Clerical Script) to simulate paleographers’ practice of leveraging transitional scripts for OBI decipherment. Concretely, our network employs a siamese framework to extract deep features from character images of various fonts, fully exploring structure clues with different resolutions by multiscale feature integration (MFI) module and multiscale refinement classifier (MRC). Experiments on cross-font dataset show CFIRN achieves state-of-the-art performance, while qualitative results on real-world undeciphered OBI data demonstrate its ability to bridge glyph evolution gaps through intermediate script associations, offering interpretable pathways for paleographic research.