Image Demosaicing
Image demosaicing reconstructs a full-color image from the incomplete data captured by a camera's color filter array (CFA) sensor. Current research emphasizes developing efficient and accurate demosaicing methods, particularly focusing on integrating demosaicing with denoising, handling various CFA patterns (including non-Bayer), and leveraging deep learning architectures like convolutional neural networks (CNNs), transformers, and neural fields. These advancements are crucial for improving image quality in various applications, from mobile photography and low-light imaging to hyperspectral imaging for medical and other fields. The development of unsupervised and efficient methods is a key focus to enable real-time processing on resource-constrained devices.
Papers
MSFA-Frequency-Aware Transformer for Hyperspectral Images Demosaicing
Haijin Zeng, Kai Feng, Shaoguang Huang, Jiezhang Cao, Yongyong Chen, Hongyan Zhang, Hiep Luong, Wilfried Philips
Inheriting Bayer's Legacy-Joint Remosaicing and Denoising for Quad Bayer Image Sensor
Haijin Zeng, Kai Feng, Jiezhang Cao, Shaoguang Huang, Yongqiang Zhao, Hiep Luong, Jan Aelterman, Wilfried Philips