Kai Feng

Kai Feng

Auto-generated profile, based on publicly available information

Kai Feng

School of Automation, Northwestern Polytechnical University, Xian, China

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 August 2021 in IEEE Transactions on Computational Imaging
Verified
Journal Article
Closed access
KFKai Feng
Kai Feng
Yong-Qiang ZhaoYong-Qiang Zhao
Yong-Qiang Zhao
Jonathan ChanJonathan Chan
Jonathan Chan
Seong G. KongSeong G. Kong
Seong G. Kong
XZXun Zhang
Xun Zhang
BWBinglu Wang
Binglu Wang
This paper presents a mosaic convolution-attention network (MCAN) for demosaicing spectral mosaic images captured using multispectral filter array (MSFA) imaging sensors. MSFA-based multispectral imaging systems acquire multispectral information of a scene in a single snap-shot operation. A complete multispectral image is reconstructed by demosaicing an MSFA-based spectral mosaic image. To avoid aliasing and artifacts in demosaicing, we utilize joint spatial-spectral correlation in a raw mosaic image. The proposed MCAN includes a mosaic convolution module (MCM) and a mosaic attention module (MAM). The MCM extracts features via a learning approach with a margin between splitting the periodic spectral mosaic and keeping the underlying spatial information of the raw image. Based on the strategy of position-sensitive weight sharing, MCM assigns the same weight to pixels with the same relative position in an MSFA. The MAM uses a position-sensitive feature aggregation strategy to describe the loading of mosaic patterns within the feature maps, which gradually reduces mosaic distortion through the attention mechanism. The experimental results on synthetic as well as real-world data show that the proposed scheme outperforms state-of-the-art methods in terms of spatial details and spectral fidelity.

22 February 2024 in IEEE Transactions on Image Processing
Verified
Journal Article
Closed access
KFKai Feng
Kai Feng
HZHaijin Zeng
Haijin Zeng
Yong-Qiang ZhaoYong-Qiang Zhao
Yong-Qiang Zhao
Seong G. KongSeong G. Kong
Seong G. Kong
YBYuanyang Bu
Yuanyang Bu
This paper presents a deep learning-based spectral demosaicing technique trained in an unsupervised manner. Many existing deep learning-based techniques relying on supervised learning with synthetic images, often underperform on real-world images, especially as the number of spectral bands increases. This paper presents a comprehensive unsupervised spectral demosaicing (USD) framework based on the characteristics of spectral mosaic images. This framework encompasses a training method, model structure, transformation strategy, and a well-fitted model selection strategy. To enable the network to dynamically model spectral correlation while maintaining a compact parameter space, we reduce the complexity and parameters of the spectral attention module. This is achieved by dividing the spectral attention tensor into spectral attention matrices in the spatial dimension and spectral attention vector in the channel dimension. This paper also presents Mosaic 25, a real 25-band hyperspectral mosaic image dataset featuring various objects, illuminations, and materials for benchmarking purposes. Extensive experiments on both synthetic and real-world datasets demonstrate that the proposed method outperforms conventional unsupervised methods in terms of spatial distortion suppression, spectral fidelity, robustness, and computational cost. Our code and dataset are publicly available at https://github.com/polwork/Unsupervised-Spectral-Demosaicing.

13 May 2024 in IEEE Transactions on Circuits and Systems for Video Technology
Verified
Journal Article
Closed access
PLPan Liu
Pan Liu
Yong-Qiang ZhaoYong-Qiang Zhao
Yong-Qiang Zhao
KFKai Feng
Kai Feng
Seong G. KongSeong G. Kong
Seong G. Kong
This paper presents a hyperspectral image (HSI) reconstruction technique based on physics-driven optimization of multispectral filter array (MSFA) patterns. The encoding of HSIs using an MSFA and their decoding through deep learning has gained increasing attention. However, previous studies have seldom explored pattern optimization from a physical perspective during the encoding process. In this paper, we apply a spectral sensitivity function (SSF) response model to generate the MSFA, and the goal of encoder optimization extends from SSF to physical structural parameters. To fully utilize spatial and spectral information in the decoding process, we design an end-to-end dual-branch spatial-spectral fusion network (DSFNet). By jointly optimizing the MSFA with the SSF response model and DSFNet, the proposed method significantly improves the reconstruction accuracy of HSI. When compared with existing HSI reconstruction methods, our proposed approach achieves state-of-the-art performance in both metric and visual quality.

20 November 2023 in Optics and Lasers in Engineering
Verified
Journal Article
Closed access
PLPan Liu
Pan Liu
YZYongqiang Zhao
Yongqiang Zhao
NLNing Li
Ning Li
KFKai Feng
Kai Feng
Seong G. KongSeong G. Kong
Seong G. Kong
CTChaolong Tang
Chaolong Tang

20 February 2025 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
HZHaijin Zeng
Haijin Zeng
KFKai Feng
Kai Feng
XZXudong Zhao
Xudong Zhao
JCJiezhang Cao
Jiezhang Cao
SHShaoguang Huang
Shaoguang Huang
WPWilfried Philips
Wilfried Philips
Hyperspectral Images (HSIs) play a pivotal role in fields like medical diagnosis and agriculture. However, it often contends with significant noise stemming from narrow-band spectral filtering. Existing denoising techniques have their limitations: model-driven methods rely on manual priors and hyperparameters, while learning-based methods struggle to discern intrinsic noise patterns, as they require paired images with specific example noise for training, fail to capture critical noise distribution information, leading to unrobust denoising results. This work addresses the issue by presenting a Degradation-Noise-Aware Unfolding Network (DNA-Net). Unlike training directly with simulated noise, DNA-Net initially models general sparse and Gaussian noise through statistic distributions. It then explicitly represents image priors with a customized spectral transformer. The model is subsequently unfolded into an end-to-end network, with hyperparameters adaptively estimated from noisy HSI and degradation models, effectively regulating each iteration. Furthermore, a novel U-Shaped Local-Non-local-Spectral Transformer (U-LNSA) is introduced, simultaneously capturing spectral correlations, local features, and non-local dependencies. The integration of U-LNSA into DNA-Net establishes the first Transformer-based deep unfolding method for HSI denoising. Experimental results on synthetic and real noise validate DNA-Net’s superior performance over state-of-the-art (SOTA) methods. Moreover, the DNA-Net, trained exclusively on mixed Gaussian noise and impulse noise, demonstrates the ability to generalize to unseen noise present in real images. Code and models will be released at https://github.com/NavyZeng/DNA-Net.

4 March 2025 in arXiv
Verified
Preprint
Open access
PDF file
JLJiahui Luo
Jiahui Luo
KFKai Feng
Kai Feng
HZHaijin Zeng
Haijin Zeng
YCYongyong Chen
Yongyong Chen
As a crucial part of the spectral filter array (SFA)-based multispectral imaging process, spectral demosaicing has exploded with the proliferation of deep learning techniques. However, (1) bothering by the difficulty of capturing corresponding labels for real data or simulating the practical spectral imaging process, end-to-end networks trained in a supervised manner using simulated data often perform poorly on real data. (2) cross-camera spectral discrepancies make it difficult to apply pre-trained models to new cameras. (3) existing demosaicing networks are prone to introducing visual artifacts on hard cases due to the interpolation of unknown values. To address these issues, we propose a hybrid supervised training method with the assistance of the self-supervised generative model, which performs well on real data across different spectral cameras. Specifically, our approach consists of three steps: (1) Pre-Training step: training the end-to-end neural network on a large amount of simulated data; (2) Pseudo-Pairing step: generating pseudo-labels of real target data using the self-supervised generative model; (3) Fine-Tuning step: fine-tuning the pre-trained model on the pseudo data pairs obtained in (2). To alleviate artifacts, we propose a frequency-domain hard patch selection method that identifies artifact-prone regions by analyzing spectral discrepancies using Fourier transform and filtering techniques, allowing targeted fine-tuning to enhance demosaicing performance. Finally, we propose UniSpecTest, a real-world multispectral mosaic image dataset for testing. Ablation experiments have demonstrated the effectiveness of each training step, and extensive experiments on both synthetic and real datasets show that our method achieves significant performance gains compared to state-of-the-art techniques.