Research and application of ceramic identification and traceability methods based on SIFT and blockchain

To address the challenges in ceramic artwork authentication and provenance—such as reduced matching precision caused by edge noise and outliers and the high storage cost of blockchain systems—we propose a hybrid SIFT (Scale Invariant Feature Transform)-Canny-Adaptive Locally-Affine Matching (AdaLAM) algorithm integrated with an Inter Planetary File System (IPFS)-blockchain solution. The proposed method first applies Canny edge detection to remove irrelevant edge points and then leverages AdaLAM’s local geometric consistency verification to dynamically optimize the matching threshold, enabling high-precision feature matching on low-texture microscopic images. On the micro-ceramic dataset, the proposed method achieves an average precision of 99.33% and an Root Mean Square Error (RMSE) of 0.0934 pixels, significantly outperforming the traditional SIFT+BF (Brute Force Matching) method (precision = 36.29%, RMSE = 32.7237 pixels), with a 63.04% improvement in precision and a 350-fold reduction in RMSE. The method maintains state-of-the-art robustness under varying scale, rotation (45°–270°), and illumination conditions (±15°), achieving a precision of 1.0. Additionally, microscopic images are stored using IPFS, resulting in a 99.9969% reduction in storage cost. A decentralized ceramic traceability system is implemented via smart contracts, offering a low-cost and highly robust solution. This work offers a novel technological pathway for ceramic identification and provenance.