Article
Mathematics
Ganbayar Batchuluun, Se Hyun Nam, Kang Ryoung Park
Summary: This study focuses on overcoming the limitations of visible light cameras and thermal cameras in plant studies based on thermal images. By using thermal images and corresponding visible light images to extract features, the accuracy of multi-class classification is improved, and a new database is built.
Article
Multidisciplinary Sciences
Salih Yanikgonul, Victor Leong, Jun Rong Ong, Ting Hu, Shawn Yohanes Siew, Ching Eng Png, Leonid Krivitsky
Summary: Integrated photodetectors are crucial components for scalable photonics platforms, with most efforts focused on devices operating at infrared telecommunication wavelengths. The authors present the first monolithically integrated avalanche photodetector for visible light, demonstrating high gain-bandwidth product and low dark current, as well as open eye diagrams at speeds up to 56 Gbps.
NATURE COMMUNICATIONS
(2021)
Article
Engineering, Multidisciplinary
Linlu Dong, Jun Wang
Summary: This study proposes a novel fusion framework called FusionPID, based on a control system, for fusing infrared and visible images. The framework utilizes a proportional integral differential (PID) control system to guide the fusion process and maintain both the thermal radiation and texture in the fused image. Experiments show that FusionPID outperforms other advanced methods in maintaining significant contrast and rich texture, and can improve the performance of downstream target detection tasks.
Article
Engineering, Electrical & Electronic
Kosom Chaitavon, Sarun Sumriddetchkajorn, Chakkrit Kamtongdee, Sataporn Chanhorm
Summary: The study introduced an optical sensing system called "Silk Check" for classifying hand reeled Thai silk yarns. The system can determine the length and linear mass density of the silk yarn in real time, and classify it correctly into corresponding grades under the Thai Agricultural Standard. Results showed that "Silk Check" can accurately classify silk yarns of different grades, benefiting the sericulture industry.
IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS
(2021)
Article
Robotics
Narayan Gatkal, Tushar Dhar, Athira Prasad, Ranganath Prajwal, Bikram Santosh, Bikram Jyoti, Ajay Kumar Roul, Rahul Potdar, Aman Mahore, Bhupendra Singh Parmar, Vala Vimalsinh
Summary: Plant phenotyping is the study of quantifying various aspects of crop plants such as quality, photosynthesis, development, growth, and biomass productivity. In the past, methods such as grid count and regression models were used, but they had limitations in terms of labor-intensity, time-consumption, and accuracy. To overcome these challenges, a portable automatic platform was developed for precise ground-based imaging. The results obtained from the image processing technique showed a high correlation with the regression model, indicating the feasibility of this approach in determining plant phenotypes.
JOURNAL OF FIELD ROBOTICS
(2023)
Review
Environmental Sciences
Ting Wen, Jian-Hong Li, Qi Wang, Yang-Yang Gao, Ge-Fei Hao, Bao-An Song
Summary: Plant phenotyping is crucial for plants to adapt to environmental changes and maintain their health. Imaging techniques, especially thermal imaging, are regarded as the most critical and reliable tools for studying plant phenotypes. This review summarizes the progress and future prospects of thermal imaging in assessing plant growth and stress responses.
SCIENCE OF THE TOTAL ENVIRONMENT
(2023)
Article
Engineering, Electrical & Electronic
Amrute Chore, Dolly Thankachan
Summary: The proposed nonintrusive multifrequency visible light analysis framework is able to accurately identify multiple nutrient deficiencies in various plants. It maps spatial and temporal light properties with nutrient shortages through extensive learning, and authenticates the readings using a deep learning technique. The model achieves high accuracy and performance in the detection of different nutrients.
JOURNAL OF ELECTRICAL ENGINEERING & TECHNOLOGY
(2023)
Article
Horticulture
Marius Ruett, Laura Verena Junker-Frohn, Bastian Siegmann, Jan Ellenberger, Hannah Jaenicke, Cory Whitney, Eike Luedeling, Peter Tiede-Arlt, Uwe Rascher
Summary: This study introduces a new method for assessing the health status of ornamental plants, using hyperspectral imaging technology combined with expert experience for plant performance monitoring. Reflectance in the green and red-edge regions of the spectrum was identified as crucial for classifying plants as healthy or stressed.
SCIENTIA HORTICULTURAE
(2022)
Review
Plant Sciences
Rijad Saric, Viet D. Nguyen, Timothy Burge, Oliver Berkowitz, Martin Trtilek, James Whelan, Mathew G. Lewsey, Edhem Custovic
Summary: Our ability to manipulate the genome exceeds our capacity to measure genetic changes on plant traits. Plant scientists have been using imaging approaches, specifically hyperspectral imaging, to define plant responses to environmental conditions and optimize crop management.
TRENDS IN PLANT SCIENCE
(2022)
Article
Optics
Chijun Li, Bin Chen, Ziliang Uan, Haoyuan Wu, Yujun Zhou, Jie Liu, Pengxin Chen, Kaixuan Chen, Changjian Guo, Liu Liu
Summary: In this study, we experimentally demonstrate an integrated visible light modulator at 532 nm on the thin-film lithium niobate platform. This modulator features low loss, large bandwidth, and provides a compact and efficient solution for visible wavelength modulation.
Article
Agriculture, Multidisciplinary
Ahmed Islam ElManawy, Dawei Sun, Alwaseela Abdalla, Yueming Zhu, Haiyan Cen
Summary: Hyperspectral imaging is a popular technique for plant phenotyping, but extracting useful traits from the images is challenging. This study introduces HSI-PP, a standalone software platform that can process and analyze hyperspectral images for high-throughput plant phenotyping. The results demonstrate its efficiency and accuracy in extracting phenotypic traits from large image datasets.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Pharmacology & Pharmacy
Marieke E. Klijn, Juergen Hubbuch
Summary: Imaging technology is increasingly utilized in biopharmaceutical formulation research, providing unique information for protein-based biopharmaceutical characterization and development studies. This review offers a comprehensive overview of analytical imaging techniques in the ultraviolet, visible, and infrared sections of the electromagnetic spectrum, outlining their applications in various (bio)physical properties and providing a future perspective on imaging in biopharmaceutical formulation research.
EUROPEAN JOURNAL OF PHARMACEUTICS AND BIOPHARMACEUTICS
(2021)
Review
Environmental Sciences
Monica Pineda, Matilde Baron, Maria-Luisa Perez-Bueno
Summary: In recent years, significant efforts have been made to develop new methods for optimizing stress detection in crop fields, with plant phenotyping based on imaging techniques becoming an essential tool in agriculture. Thermal imaging, particularly leaf temperature, is a valuable indicator of plant physiological status responsive to both biotic and abiotic stressors. When combined with other imaging sensors and data-mining techniques, thermography plays a crucial role in achieving more automated, precise, and sustainable agriculture.
Article
Agronomy
Huichun Zhang, Lu Wang, Xiuliang Jin, Liming Bian, Yufeng Ge
Summary: Acquisition of plant phenotypic information is crucial for plant breeding and gene regulation, as well as optimization of agricultural and forestry product quality. Optical sensors and data processing methods enable measurement of leaf morphological, physiological, and biochemical traits at various levels, facilitating rapid and accurate acquisition of plant leaf phenotypes.
Article
Mathematics
Ganbayar Batchuluun, Se Hyun Nam, Chanhum Park, Kang Ryoung Park
Summary: This study proposes a novel plant classification method based on both thermal and visible-light images, which shows higher accuracies than existing methods. It is the first study to perform super-resolution reconstruction using visible-light and thermal plant images, and a method to improve classification performance is proposed using generative adversarial network (GAN)-based super-resolution reconstruction.
Article
Anatomy & Morphology
Tianyu Hu, Xiaofeng Xu, Shangbin Chen, Qian Liu
Summary: The study introduces a novel neuronal soma segmentation method that combines 3D U-shaped fully convolutional neural networks with multi-task learning, aiming to address the challenges posed by touching neuronal somata and variable soma shapes in images. This technique outperforms existing methods by applying multi-task learning to predict soma boundaries for splitting touching somata and utilizing a U-shaped convolutional neural network effective for limited datasets. The proposed method also incorporates a contour-aware multi-task learning framework and a spatial attention module to simultaneously predict masks of neuronal somata and boundaries, resulting in improved segmentation results for high-throughput analysis of large-scale optical imaging data.
FRONTIERS IN NEUROANATOMY
(2021)
Review
Biochemistry & Molecular Biology
Xingzhou Peng, Junjie Wang, Feifan Zhou, Qian Liu, Zhihong Zhang
Summary: Immunotherapies are safe and effective for tumor treatments, with the lymphatic system playing a crucial role in modulating the immune response. Targeting the lymphatic system using nanoparticle technology shows promise in improving the efficacy of cancer treatments.
CELLULAR AND MOLECULAR LIFE SCIENCES
(2021)
Article
Optics
Laixiang Xu, Fuhong Cai, Yuxin Hu, Zhen Lin, Qian Liu
Summary: The study converted spectral data into two-dimensional matrix data and processed it using deep learning algorithms to accurately classify reflectance spectra. The algorithm achieved a maximum classification accuracy of 99.56% and an average accuracy of 96.78%. This method enables the analysis of spectral data with deep learning algorithms, allowing for more accurate analysis of complex spectral data in the future.
Article
Plant Sciences
Keith E. Duncan, Kirk J. Czymmek, Ni Jiang, August C. Thies, Christopher N. Topp
Summary: This study presents technical advances in using lab-based X-ray microscopy for high-resolution 3D imaging of plant samples at multiple scales. Serial imaging and improved sample preparation methods allow for the generation of sub-micron 3D volumes co-registered with lower magnification scans, providing explicit contextual reference. This method bridges the imaging gap between light and electron microscopy and can be applied to various economically and scientifically important plant systems.
Article
Biochemical Research Methods
Xiangjiang Dong, Ling Fu, Qian Liu
Summary: This research proposes a no-reference image quality assessment method for confocal endoscopy images based on Weber's law and local descriptors, which can effectively detect the severity of image degradation.
JOURNAL OF BIOMEDICAL OPTICS
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Yulin Wang, Wenyuan Wu, Yuxin Yang, Haifeng Hu, Shangqian Yu, Xiangjiang Dong, Feng Chen, Qian Liu
Summary: This study proposed a deep learning framework to synthesize 3D full-contrast MR images from 2D images, addressing the issues caused by GBCAs injection. The improved network showed excellent performance in quantitative and qualitative evaluations, instilling high confidence in diagnosis.
Article
Computer Science, Interdisciplinary Applications
Qiang Zeng, Gang Zheng, Qian Liu
Summary: This paper proposes a method called PE-DLS for full-body motion reconstruction in two stages: pose estimation (PE) and damped least squares (DLS) optimization. The method outperforms others in terms of reconstruction error and computational time, and a full-body MoCap system based on Vive devices is implemented.
Article
Computer Science, Interdisciplinary Applications
Laixiang Xu, Fuhong Cai, Yanhu Fu, Qian Liu
Summary: Cervical cancer poses a serious threat to women's lives and health. In order to improve the recognition accuracy of cervical cell smear images, we propose a novel deep-learning model based on improved Faster R-CNN, shallow feature enhancement networks, and generative adversarial networks. Experimental results show that our methods outperform other algorithms in terms of shorter time consumption, higher recognition precision, and stronger adaptive ability. Our method provides a useful reference for cervical cell smear image analysis.
MEDICAL & BIOLOGICAL ENGINEERING & COMPUTING
(2023)
Article
Biochemical Research Methods
Jingjun Zhou, Xiangjiang Dong, Qian Liu
Summary: This paper proposes a feature-level MixSiam method based on the traditional Siamese network for learning the discriminative features of probe-based confocal laser endomicroscopy (pCLE) images for gastrointestinal (GI) tumor classification. The method consists of two stages: self-supervised learning (SSL) and few-shot learning (FS). In the SSL stage, a feature mixing approach is introduced to enhance the adaptation of the Siamese structure to the intra-class variance of the pCLE dataset. In the FS stage, a pre-trained model obtained through SSL is used as the base learner to enable rapid generalization to other pCLE classification tasks with limited labeled data. Experimental results demonstrate the effectiveness of the proposed method in improving the classification of pCLE images for different stages of tumor development.
BIOMEDICAL OPTICS EXPRESS
(2023)
Article
Plant Sciences
Yuwei Lu, Rui Wang, Tianyu Hu, Qiang He, Zhou Shuai Chen, Jinhu Wang, Lingbo Liu, Chuanying Fang, Jie Luo, Ling Fu, Lejun Yu, Qian Liu
Summary: This study developed a Micro-CT system for automated nondestructive imaging and 3D modeling of passion fruit samples. Deep learning models were used for segmentation and label generation, enabling the automatic calculation of 14 traits. The results show that the measurements of external traits are comparable to manual operations, while the measurements of internal traits are more reliable. The study also identified correlations among the traits and proposed a scoring method for comprehensive quality evaluation.
FRONTIERS IN PLANT SCIENCE
(2023)
Article
Plant Sciences
Lejun Yu, Lingbo Liu, Wanneng Yang, Dan Wu, Jinhu Wang, Qiang He, ZhouShuai Chen, Qian Liu
Summary: With the use of the Micro-CT system and deep learning models, automated, accurate, and nondestructive measurements of coconut fruits and seeds can be achieved, enabling the acquisition of a wealth of agronomic and digital traits information.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Environmental Sciences
Dong Xu, Yuwei Lu, Heng Liang, Zhen Lu, Lejun Yu, Qian Liu
Summary: The areca nut is an important economic source in southeast Asia, but the emergence of areca yellow leaf disease has greatly reduced its production. This study proposes a high-precision method to predict the severity of the disease using unmanned aerial vehicles and machine learning algorithms. The correlation between canopy temperature and the disease is also demonstrated.
Article
Engineering, Electrical & Electronic
Yunfei Li, Fuzhou Shen, Lantian Hu, Ziyue Lang, Qian Liu, Fuhong Cai, Ling Fu
Summary: This article proposes a novel high-speed hyperspectral imaging (HSI) system which enables real-time monitoring of dynamic biological samples with high spectral and spatial resolution. The system utilizes a high-speed galvo mirror and a 10-Gb ethernet port CMOS sensor for spatial scanning and data acquisition, achieving a high imaging speed without sacrificing resolution.
IEEE SENSORS JOURNAL
(2023)
Article
Physics, Applied
Jing Cao, Ling Fu, Pinghe Wang, Qian Liu
Summary: This study proposes a unique method to determine the actual resolution at each imaging point using reflection matrix optical coherence tomography (RMOCT). It also introduces the concept of contribution rate to quantitatively evaluate the imaging quality. This study represents a comprehensive assessment of the practical performance of RMOCT in terms of actual resolving power and imaging quality.
JOURNAL OF APPLIED PHYSICS
(2023)
Article
Agronomy
Dan Wu, Lejun Yu, Junli Ye, Ruifang Zhai, Lingfeng Duan, Lingbo Liu, Nai Wu, Zedong Geng, Jingbo Fu, Chenglong Huang, Shangbin Chen, Qian Liu, Wanneng Yang
Summary: This research presents an automatic and nondestructive method for 3D panicle modeling in rice plants. The method integrates various techniques such as shoot rice reconstruction, shape from silhouette, deep convolutional neural network, ray tracing, and supervoxel clustering. It demonstrates high efficiency and performance in recovering the 3D shapes of rice panicles from multiview images, and is adaptable to diverse rice plants. The proposed algorithm outperforms the classical structure-from-motion method in terms of texture preservation and computational efficiency.