Article
Engineering, Electrical & Electronic
Lingyan Ruan, Bin Chen, Jizhou Li, Miu-Ling Lam
Summary: The study proposes a novel convolutional neural network architecture AIFNet for removing spatially-varying defocus blur from a single defocused image. It combines light field synthetic aperture and refocusing techniques and demonstrates state-of-the-art performance in both quantitative and qualitative evaluations.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2021)
Article
Computer Science, Artificial Intelligence
Yaowei Li, Jinshan Pan, Ye Luo, Jianwei Lu
Summary: This paper proposes an exemplar-based method to solve the problem of dynamic scene deblurring, using siamese encoder network and shallow encoder network to extract features and develop a rank module to explore useful features, achieving significant improvements in both quantitative and qualitative evaluations.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Y. Demir, N. H. Kaplan
Summary: Low-light images suffer from poor visibility, severe noise, low contrast, and low brightness. This paper proposes a low-light image enhancement method that applies the HSV transform and a multi-scale decomposition of the SSIF to obtain approximation and detail sub-images of the V component. Contrast-Limited Adaptive Histogram Equalization (CLAHE) is applied to the approximation image for higher contrast, and the amplified detail sub-images are added to reconstruct the enhanced V component. The results show that this method provides better visual quality and more natural colors than state-of-the-art methods.
DIGITAL SIGNAL PROCESSING
(2023)
Article
Engineering, Electrical & Electronic
Meng Chang, Chenwei Yang, Huajun Feng, Zhihai Xu, Qi Li
Summary: A novel method for handling camera motion blur with outliers is proposed in this paper, which utilizes an edge-aware scale-recurrent network and a salient edge detection network for deblurring and edge restoration constraint. Experimental results demonstrate that this method outperforms other existing approaches in terms of deblurring quality.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2021)
Article
Computer Science, Artificial Intelligence
Xiao Zhou, Xiaobiao Du, Peizhe Ru
Summary: This paper presents a low-light image enhancement method based on deep learning to improve autonomous piloting tasks. The lack of sufficient photon signals in low-light environments affects the performance of vision-based autonomous piloting. The proposed method utilizes synthesized noisy-clean data pairs to train a neural network that enhances the quality of dark light images, improving downstream auto-piloting applications.
IET IMAGE PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Fu-Jen Tsai, Yan-Tsung Peng, Chung-Chi Tsai, Yen-Yu Lin, Chia-Wen Lin
Summary: This paper proposes a Blur-aware Attention Network (BANet) for accurate and efficient deblurring through a single forward pass, utilizing region-based self-attention with multi-kernel strip pooling to disentangle blur patterns of different magnitudes and orientations, and cascaded parallel dilated convolution to aggregate multi-scale content features. Extensive experimental results show that BANet outperforms state-of-the-arts in blurred image restoration and can provide real-time deblurred results.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Computer Science, Information Systems
Shunsuke Yae, Masaaki Ikehara
Summary: With the increasing use of smartphones and digital video cameras, the need to handle digital video has become more significant. This study proposes a software-based method to deblur videos by processing them after shooting. By constructing a multi-scale network based on UNet and making improvements based on the structure of MPRNet, the proposed method achieved impressive results in deblurring videos.
Article
Computer Science, Software Engineering
Tianlin Zhang, Jinjiang Li, Hui Fan
Summary: Researchers have proposed a new method for deblurring images of dynamic scenes by improving the network structure and scale perception design. This method can handle changes in blurring at different scales and significantly improve image restoration and texture details.
COMPUTATIONAL VISUAL MEDIA
(2022)
Article
Environmental Sciences
Ziyu Zhang, Liangliang Zheng, Yongjie Piao, Shuping Tao, Wei Xu, Tan Gao, Xiaobin Wu
Summary: In this paper, an algorithm based on local binary pattern (LBP) is proposed to obtain clear remote sensing images under unknown causes of blurring. Experiments show that the proposed method achieves highly competitive results in remote sensing image processing.
Article
Computer Science, Information Systems
Ce Wang, Dejia Xu, Renjie Wan, Bin He, Boxin Shi, Ling-Yu Duan
Summary: Colored glass often degrades images with reflection and color bias, making the recovery of the clean background challenging. In this paper, we propose a cooperative framework to address the mutual interference problem caused by colored glass, along with a novel glass color invariant loss and progressive refinement. Experimental results demonstrate the state-of-the-art performance of our method.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Computer Science, Artificial Intelligence
Hayat Ullah, Khan Muhammad, Muhammad Irfan, Saeed Anwar, Muhammad Sajjad, Ali Shariq Imran, Victor Hugo C. de Albuquerque
Summary: This study introduces a lightweight convolutional neural network, LD-Net, for reconstructing hazy images and proposes a color visibility restoration method to overcome weather challenges. Extensive experiments validate the superiority of the proposed method in image dehazing techniques and object detection tasks.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Hyungjoo Jung, Youngjung Kim, Hyunsung Jang, Namkoo Ha, Kwanghoon Sohn
Summary: By considering the relationship between motion and blur, a motion-aware feature learning framework for dynamic scene deblurring through multi-task learning has been proposed. The framework simultaneously estimates a deblurred image and a motion field from a blurred image, showing improved performance in image deblurring and motion estimation tasks. Additional analysis and experimental results demonstrate the effectiveness of the proposed method in outperforming state-of-the-art deblurring techniques.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Kaihao Zhang, Wenqi Ren, Wenhan Luo, Wei-Sheng Lai, Bjorn Stenger, Ming-Hsuan Yang, Hongdong Li
Summary: This paper presents a comprehensive survey of recently published deep-learning based image deblurring approaches. The paper introduces a taxonomy of methods using convolutional neural networks (CNN) based on architecture, loss function, and application, and discusses some domain-specific deblurring applications. The paper concludes by discussing key challenges and future research directions.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2022)
Article
Computer Science, Information Systems
Chih-Hung Liang, Yu-An Chen, Yueh-Cheng Liu, Winston H. Hsu
Summary: This article presents a deep learning-based blind image deblurring method, which achieves state-of-the-art performance by training on RAW images. The authors also create a new dataset containing RAW images and processed sRGB images, and demonstrate that training on this dataset can improve existing deblurring models.
IEEE TRANSACTIONS ON MULTIMEDIA
(2022)
Article
Computer Science, Artificial Intelligence
Salman Siddique Khan, Varun Sundar, Vivek Boominathan, Ashok Veeraraghavan, Kaushik Mitra
Summary: Lensless imaging is a potential solution for ultra-miniature cameras, relying on computational algorithms to recover scenes from measurements. This study proposes a non-iterative deep learning approach called FlatNet, which significantly improves image quality for lensless reconstructions. FlatNet consists of two stages trained in an end-to-end manner, producing fast and photorealistic reconstructions for mask-based lensless cameras.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Engineering, Electrical & Electronic
Daisuke Sugimura, Tomoaki Yamazaki, Takayuki Hamamoto
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2020)
Article
Computer Science, Information Systems
Masahito Shimamoto, Yusuke Kameda, Takayuki Hamamoto
IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS
(2020)
Article
Computer Science, Hardware & Architecture
Misaki Shikakura, Yusuke Kameda, Takayuki Hamamoto
Summary: This paper discusses the evolution and potential applications of image sensors equipped with high-speed brightness gradient sensors. An adaptive exposure time control method based on apparent motion estimation from the sensor is proposed and results are evaluated for changes in illuminance and global/local motion.
IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES
(2021)
Article
Computer Science, Information Systems
Katsuya Fujii, Daisuke Sugimura, Takayuki Hamamoto
Summary: The method proposed in this paper focuses on group-level emotion recognition with the contributions of hierarchical classification approach and novel features. The use of facial expressions for emotion differentiation, as well as the incorporation of scene features and object-wise semantic information, improves the performance of emotion recognition.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Computer Science, Artificial Intelligence
Kosuke Kurihara, Daisuke Sugimura, Takayuki Hamamoto
Summary: The proposed non-contact heart rate estimation method utilizes adaptive fusion of RGB and NIR signals for accurate estimation regardless of the lighting conditions. Experimental results show the effectiveness of the method in capturing heart rate in various scenarios.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Engineering, Electrical & Electronic
Kurumi Kataoka, Yusuke Kameda, Takayuki Hamamoto
Summary: The proposed adaptive exposure-time-control method for image sensors adjusts exposure time for each pixel based on scene luminance and contrast to reconstruct high-dynamic-range images while minimizing blown-out highlights and blocked-up shadows. By maximizing entropy of the entire image and for individual blocks, the method effectively estimates appropriate exposure times for different areas in the scene and reduces adjustment time by capturing and estimating multiple exposure times simultaneously. Simulation experiments demonstrate the effectiveness of the method.
ITE TRANSACTIONS ON MEDIA TECHNOLOGY AND APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Kiyotaka Iwabuchi, Yusuke Kameda, Takayuki Hamamoto
Summary: The proposed method in this article involves deblurring of object motion units in a scene through motion compensation, achieving more efficient motion blur suppression. It also applies a novel technique for accurate motion estimation from the bit-plane frame even in photon-limited situations through statistical evaluation. Experimental results show that the method can also be applied for denoising, improving the peak signal-to-noise ratio.
Proceedings Paper
Imaging Science & Photographic Technology
Shunsuke Ishihara, Kazuya Kodama, Takayuki Hamamoto
2020 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)
(2020)
Proceedings Paper
Computer Science, Interdisciplinary Applications
H. Ikeoka, T. Hamamoto
INTERNATIONAL WORKSHOP ON ADVANCED IMAGING TECHNOLOGY (IWAIT) 2020
(2020)
Proceedings Paper
Computer Science, Interdisciplinary Applications
Yusuke Kameda, Yoshihiro Maeda, Takayuki Hamamoto
INTERNATIONAL WORKSHOP ON ADVANCED IMAGING TECHNOLOGY (IWAIT) 2020
(2020)
Article
Engineering, Electrical & Electronic
Takayuki Honda, Daisuke Sugimura, Takayuki Hamamoto
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2020)
Article
Engineering, Electrical & Electronic
Mamoru Sugawara, Kazunori Uruma, Seiichiro Hangai, Takayuki Hamamoto
IEEE SIGNAL PROCESSING LETTERS
(2020)
Proceedings Paper
Engineering, Electrical & Electronic
Katsuya Fujii, Daisuke Sugimura, Takayuki Hamamoto
2019 14TH IEEE INTERNATIONAL CONFERENCE ON AUTOMATIC FACE AND GESTURE RECOGNITION (FG 2019)
(2019)
Proceedings Paper
Optics
Seiya Suda, Kazuya Kodama, Takayuki Hamamoto
INTERNATIONAL WORKSHOP ON ADVANCED IMAGE TECHNOLOGY (IWAIT) 2019
(2019)
Proceedings Paper
Optics
Hiroshi Ikeoka, Takayuki Hamamoto
INTERNATIONAL WORKSHOP ON ADVANCED IMAGE TECHNOLOGY (IWAIT) 2019
(2019)