A dual residual image restoration network for nuclear noise image denoising
Abstrak
To better preserve the texture detail information of nuclear noise images after denoising in a nuclear environment, a nuclear noise image denoising method based on a dual residual network is proposed. The proposed DRADNet consists of different branch sub-networks composed of the residual channel self-attention module (GCARB) and the multi-semantic space residual module (MS-SRB), which enhances the model’s learning ability by capturing complementary feature information of the image. Each sub-network contains five residual attention blocks, which capture multi-scale feature information of the image through sampling operations and long skip connections. The feature fusion module (FFMB) fuses the features extracted by the two branch networks, making the flat areas of the image smoother and the texture areas sharper, to obtain higher-quality and clearer images. A large number of experiments have shown that, compared with other state-of-the-art denoising methods, the denoising effect of DRADNet is the most outstanding.
Topik & Kata Kunci
Penulis (4)
Xue Gao
Zhiqiang Wu
Jie Liu
Jie Chen
Akses Cepat
- Tahun Terbit
- 2026
- Bahasa
- en
- Sumber Database
- Semantic Scholar
- DOI
- 10.1117/12.3107187
- Akses
- Open Access ✓