测试技术学报2024，Vol.38Issue(3)：281-288,8.DOI:10.3969/j.issn.1671-7449.2024037
基于空间-光谱级联差分网络的病理学高光谱图像融合方法
Pathological Hyperspectral Image Fusion Method Based on Cascaded Spatial-Spectral Differential network
摘要
Abstract
High-resolution hyperspectral pathology images contain fine-grained two-dimensional spatial information and spectral dimensional information,which is of great significance for accurate diagnosis.To this end,we propose a pathological hyperspectral image fusion method based on the space-spectrum cascade differential network.In this method,high-resolution hyperspectral pathological images are reconstructed by cascading spatial-spectral modules.Each space-spectrum module utilizes differential methods to design spatial edge loss and spectral edge loss,which helps constrain and optimize the model in stages.This enables the organic fusion of pathological hyperspectral images and multispectral images.The model experimented on the pathological image dataset,and the four evaluation indexes RMSE,PSNR,ERGAS,and SAM of the fusion images reached 4.593 7,32.328 0,4.668 3 and 3.635 4,respectively.These results demonstrate that the fusion method based on the space-spectrum cascade differential network can achieve fine fusion of pathological images with rich details,providing a reference for multispectral image fusion.关键词
卷积神经网络/级联差分/空间边缘损失/光谱边缘损失/病理学图像融合Key words
convolutional neural network/cascade difference/spatial edge loss/spectral edge loss/pathological image fusion分类
信息技术与安全科学引用本文复制引用
王朝亮,梁美彦..基于空间-光谱级联差分网络的病理学高光谱图像融合方法[J].测试技术学报,2024,38(3):281-288,8.基金项目
山西省回国留学人员科研资助项目(2023-010) （2023-010）
国家自然科学基金资助项目(11804209) （11804209）
