.png){kind=icon}
返回
ImpRes: implicit residual diffusion models for image super-resolution
.png){kind=icon}
PRE
AI
DOI：10.1007/s00371-024-03717-4.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Single-image super-resolution (SISR) is a fundamental task in computer vision that faces challenges due to the loss of high-frequency information during image degradation, leading to a nonunique solution space. Current super-resolution (SR) methods often suffer from high-frequency texture distortion, excessive smoothing, and scale inconsistency. This study introduces an innovative implicit residual diffusion model (ImpRes) to address these issues. ImpRes enhances model convergence speed and high-frequency detail recovery through a residual prediction mechanism. It incorporates a Gaussian frequency decomposition module using Gaussian high-pass filters to emphasize high-frequency components, guiding accurate texture reconstruction. Additionally, ImpRes employs static implicit neural representation (SINR) during decoding to transform discrete image representations into a continuous local implicit image function, achieving precise content perception, flexible spatial sampling, and mitigating over-smoothing. Experimental results demonstrate that ImpRes outperforms most existing diffusion-based methods in terms of model convergence time, generation quality, and scale consistency, achieving a peak signal-to-noise ratio of 29.97 dB in 4 x face super-resolution tasks. Our implementation is available at: https://github.com/fineverse/ImpRes.
Keyword:
Diffusion models
Image super-resolution
Residual prediction
Implicit neural representation
期刊
IF:
2.9
论文数:
4.6K
被引数:
6.5K
机构
引用论文
SRDiff: Single image super-resolution with diffusion probabilistic modelsSRDiff: 具有扩散概率模型的单幅图像超分辨率.png)
NEUROCOMPUTING
OAAI
NEUROCOMPUTINGIF6.5
OAAI分享收藏n-CD: A geometric approach to preserving location privacy in location-based servicesN-cd: 一种在基于位置的服务中保护位置隐私的几何方法
PREAI
分享收藏分享收藏Status of Malaria Infection in Peri-Urban Community of North Central Region of Nigeria尼日利亚中北部地区城郊社区疟疾感染状况OAAI
OAAI分享收藏PointALCR: adversarial latent GAN and contrastive regularization for point cloud completionVISUAL COMPUTER
VISUAL COMPUTERIF2.9
PREAI
分享收藏Hysterectomy Trends: A Canadian Perspective on the Past, Present, and Future子宫切除术趋势：加拿大的过去、现在与未来视角OAAI
OAAI分享收藏“Nevertheless, it is still painful”: Use of Over-the-Counter analgesics, pain, stress, and HRQOL in Norwegian adolescents: a cross sectional study.“然而，这仍然令人痛苦”：非处方镇痛药的使用、疼痛、压力与挪威青少年健康相关生命质量：一项横断面研究。
OAAI
IF0
OAAI分享收藏
查看更多内容