Image fusion method based on RFN-Nest

• Main Authors: WANG XIAOHUA, PAN XIAOGUANG, JIAO LULU, ZHANG YANA, ZHANG NA
• Format: Patent
• Language: Chinese; English
• Published: 12.07.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention belongs to the technical field of machine learning, and particularly relates to an RFN-Nest-based image fusion method, which comprises the following steps: S1, data acquisition: collecting a common data set for infrared and visible light images, and completing data construction required by model training; s2, data preprocessing: preprocessing comprises normalization, data division and data scale unification, the preprocessed image is input into the constructed model, and the model training effect is ensured; s3, identifying the model: completing model building and model training by adopting deep learning; and S4, model storage: when the loss function of the model is not reduced any more, storing the model. According to the method, the multi-scale features are extracted from the source image by using the trained encoder, and the fused image is reconstructed by using the fused multi-scale features. In the second stage, Ldetail and Lf efeature are respectively used to retain the significant feature