2026-10-14T21:35:44Z https://www.scidb.cn/oai

10.57760/sciencedb.j00186.01174 2026-07-27T19:30:54Z

2026-07-27 Selected Supporting Image Dataset for “T-DDFM: An Advanced Denoising Diffusion Fusion Model for High-Precision Neutron Imaging” doi:10.57760/sciencedb.j00186.01174 en This dataset provides selected supporting image data for the manuscript “T-DDFM: An Advanced Denoising Diffusion Fusion Model for High-Precision Neutron Imaging”. It contains 26 PNG image files, with a total size of approximately 9.13 MiB. The dataset includes representative images used for multimodal image fusion, neutron-image denoising of a nickel-based alloy turbine blade, and training-strategy comparison experiments using slotted iron and stainless-steel samples.The study focuses on improving neutron image quality under low-flux and long-exposure imaging conditions. The proposed T-DDFM method integrates tensor robust principal component analysis (TRPCA) with a denoising diffusion fusion model to suppress complex noise, including Gaussian noise, Poisson noise, and white-spot noise, while preserving structural details and improving the visibility of fine features in neutron radiography.The neutron imaging experiments described in the manuscript were conducted using a 2.5 MeV D-D neutron generator. The imaging system included a neutron source, collimator with polyethylene moderator, sample stage, ZnS/6LiF scintillator, mirror, light-tight shielding enclosure, CCD camera, and image-processing software. The estimated thermal-neutron flux at the sample position was 10^4–10^5 n/(cm²·s), and the source-to-sample distance was adjustable from 54 to 114 cm.The dataset is organized into three main parts: multimodal brain-image fusion examples, turbine-blade neutron-image denoising examples, and two-metal-sample training-variant comparison images. It also includes metadata files, a representative image, a file manifest, checksum information, a data dictionary, and a quality-control report. The image data can be used as reference material for studies on neutron image denoising, diffusion-model-based image processing, image fusion, industrial nondestructive testing, and high-precision neutron radiography.This release contains selected supporting images currently available and is not the complete 150-image neutron training dataset described in the manuscript. The complete original exposure sequences, source code, model weights, ROI coordinates, and metric-computation scripts are not included in this version. Neutron imaging; Neutron radiography; T-DDFM; Denoising diffusion model; Image denoising; Image fusion; Image enhancement 刘源煜 EMBARGO https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank