I am currently living with my parents and applying for a PhD at SCUT, Guangzhou. My research interest lies in tensor analysis, image restoration and explanable deep learning. I received the B.S. in applied mathematics and M.Sc. in software engineering from SCUT in 2016 and 2019, respectively. I enjoy studying English and I am a part-time translator. Now my priority is to try to lose some weight.

News

• 01/2022, continue the “Denoising Comparison Project”.
• 12/2021, leave Lehigh University and start to pursue a PhD at SCUT.
• 09/2021, one paper has been rejected by IEEE TMI.
• 12/2020, one paper has been rejected by AAAI 2021.
• 01/2020, one paper has been rejected by CVPR 2020.
• 09/2019, join Lehigh University as a research assistant.
• 03/2019, one paper has been accepted by IEEE TIP.
• 04/2018, one paper has been accepted by IEEE TMI.

Publications

• Kong, Zhaoming, and Xiaowei Yang. “Color image and multispectral image denoising using block diagonal representation.” IEEE Transactions on Image Processing 28.9 (2019): 4247-4259. Paper and Code

• Kong, Zhaoming, Le Han, Xiaolan Liu and Xiaowei Yang. “A New 4-D nonlocal transform-domain filter for 3-D magnetic resonance images denoising.” IEEE transactions on medical imaging 37.4 (2017): 941-954. Paper and Code

Denoising Comparison Project

In this project, we extensively compare numerous on both synthetic and real-world datasets. We also work hard to collect new color image and video datasets for benchmarking, and our evaluations are performed from different perspectives. We will continuously update our datasets and denoising results of compared methods on this page.

• The Proposed Dataset: I have spent over three years collecting image data based on real-world scenes using more than 10 different cameras. The proposed dataset is based on indoor and outdoor scenes. I do not use predefined camera settings such as ISO, shutter speed and aperture. Instead, I mainly use the cameras’ auto-mode, which I believe is the default and most widely-used mode when taking photos. Therefore, the noise levels of images in my dataset vary significantly. In my opinion, a good denoiser should be robust to noise while preserve as many details as possible.