Dunhuang book fragment conjugation method based on multi-feature mining and cooperative constraint

• Main Authors: CHEN SHIQIANG, YAO LEI, MIAO CHANGHONG, FAN GAOJUAN, HU RENFEN, LIU HAOXIA, MEN YI, ZHANG CHONGSHENG, CHEN KE, MO BOFENG
• Format: Patent
• Language: Chinese; English
• Published: 09.06.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention discloses a Dunhuang book fragment conjugation method based on multi-feature mining and cooperative constraint. The method comprises the following steps: A, preprocessing an image of a Dunhuang book fragment to be conjugated; b, constructing a multi-feature extraction network which comprises a heterogeneous feature mining module, a feature aggregation module and a feature coding module; c, optimizing the multi-feature extraction network by using a comparative learning module; and D, inputting the preprocessed image of the Dunhuang book fragments to be conjugated into the optimized multi-feature extraction network to extract key features, and judging the conjugation matching degree of the plurality of Dunhuang book fragments to be conjugated by using the encoded aggregation features. According to the method, the multi-feature extraction network is utilized to fully mine the writing style feature, the text layout feature, the residual edge feature and the statement smoothness feature of the Dunhua