Article
Engineering, Electrical & Electronic
Jiayi Ma, Linfeng Tang, Meilong Xu, Hao Zhang, Guobao Xiao
Summary: The proposed STDFusionNet is a fusion network for infrared and visible images that preserves thermal targets and texture structures. By utilizing salient target detection and a specific loss function, it successfully extracts and reconstructs features to achieve high-quality fusion results. Through qualitative and quantitative experiments, the algorithm has shown superiority in terms of speed and quality improvement over existing methods.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Chemistry, Analytical
Jun Fan, Jingbiao Wei, Hai Huang, Dafeng Zhang, Ce Chen
Summary: This study proposes a framework for infrared vehicle small target detection and tracking, which consists of three components: full-image object detection, cropped-image object detection and tracking, and object trajectory prediction. The study designs a CNN-based real-time detection model with a high recall rate for the first component, and uses the KCF algorithm and a lightweight CNN-based target detection model for the second component to lock on the target more precisely. An optimized Kalman filter is designed for the final component to estimate the target's trajectory. Validation on a public dataset shows that the proposed framework can steadily track vehicle targets and adapt well in challenging situations.
Article
Green & Sustainable Science & Technology
Honglyun Park, Jaewan Choi
Summary: The study utilized Worldview-3 satellite imagery for mineral detection, enhancing the utilization of VNIR and SWIR bands through pansharpening technique, and applying various similarity analysis techniques for mineral detection, finally selecting pixels indicating minerals through a majority voting technique.
Article
Geochemistry & Geophysics
Xiaoqing Tian, Jing Liu, Mahendra Mallick, Kaiyu Huang
Summary: This article presents a new algorithm for tracking moving targets using video synthetic-aperture radar (ViSAR) images. By utilizing strategies such as expansion and contraction, as well as region partitioning, the algorithm successfully detects and tracks multiple dim targets. Experimental results demonstrate that the proposed algorithm outperforms existing methods in terms of location accuracy and false-alarm suppression.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Engineering, Electrical & Electronic
Hadi Shahraki, Saed Moradi, Shokoufeh Aalaei
Summary: In this paper, a noise-robust method called Branch Local Contrast Measure (BLCM) is proposed for infrared small target detection. The method enhances the targets based on the differences between real targets and noise using a suitable filter. The effectiveness of the proposed method is evaluated using 431 IR images with small dim targets containing various sources of false response, and the results show that BLCM outperforms other methods for small target detection in noisy images.
SIGNAL IMAGE AND VIDEO PROCESSING
(2023)
Article
Environmental Sciences
Tobias Hupel, Peter Stuetz
Summary: This paper investigates and evaluates the applicability of four well-known hyperspectral anomaly detection methods and a new method called local point density for real-time camouflage detection in multispectral imagery. The results show that these methods have generally high detection performance on various targets.
Article
Computer Science, Artificial Intelligence
Sijie Wu, Kai Zhang, Shaoyi Li, Jie Yan
Summary: The study introduces an airborne infrared target tracking algorithm that utilizes feature embedding learning and correlation filters for improved performance. By developing a shallow network and a contrastive center loss function to learn the prototypical representation of the aircraft in the embedding space, and integrating the feature embedding module into the efficient convolution operator framework for aircraft tracking, the research achieves effective tracking of aircraft targets.
Article
Computer Science, Artificial Intelligence
Xuedong He, Calvin Yu-Chian Chen
Summary: This paper proposes a lightweight feature separation and fusion module to improve the performance of discriminative trackers. Additionally, a target uncertain detection technique is designed to address the problem of tracking model corruption. Through comprehensive experimental evaluations, the results demonstrate the excellent performance of the proposed methods on seven public benchmarks.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Environmental Sciences
Zheyong Li, Jinghua Li, Pei Zhang, Lihui Zheng, Yilong Shen, Qi Li, Xin Li, Tong Li
Summary: This paper proposes a transfer-based underwater target detection framework (TUTDF), which utilizes synthetic data to train deep-learning models and transfers them to real-world applications. The framework tackles the challenge of distribution disparity between real and synthetic data by dividing the domains using depth information. It also applies a spatial-spectral process to eliminate the adverse influence of background noise.
Article
Computer Science, Artificial Intelligence
J. F. Ciprian-Sanchez, G. Ochoa-Ruiz, M. Gonzalez-Mendoza, L. Rossi
Summary: The researchers selected three state-of-the-art deep learning-based image fusion techniques and evaluated their performance on fire image fusion task. They also proposed an improved method for generating artificial infrared and fused images.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Instruments & Instrumentation
Kun Qian, Sheng Hui Rong, Kuan Hong Cheng
Summary: This paper presents a tracking algorithm that combines KCF with a detection model to tackle challenges in infrared dual band imagery. By using SWBF to suppress background edge, computing fused features and introducing a detection model, it solves migration problem and improves accuracy and robustness.
INFRARED PHYSICS & TECHNOLOGY
(2021)
Article
Geochemistry & Geophysics
Jinmiao Zhao, Chuang Yu, Zelin Shi, Yunpeng Liu, Yingdi Zhang
Summary: In this paper, we propose an innovative gradient-guided learning network (GGL-Net) for infrared small target detection. By introducing gradient magnitude images and constructing a dual-branch feature extraction network with a two-way guidance fusion module, precise positioning of small targets and effective feature extraction are achieved.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Geochemistry & Geophysics
Jiaqing Zhang, Jie Lei, Weiying Xie, Zhenman Fang, Yunsong Li, Qian Du
Summary: In this article, the authors propose SuperYOLO, an accurate and fast object detection method for remote sensing images. By fusing multimodal data and utilizing assisted super resolution learning, SuperYOLO achieves high-resolution object detection on multiscale objects while considering the computation cost. Experimental results show that SuperYOLO outperforms state-of-the-art models in terms of accuracy and computational efficiency.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Bin Zhao, Chunping Wang, Qiang Fu, Zishuo Han
Summary: This study proposes a detection pattern based on generative adversarial network to focus on the essential features of infrared small targets, recognizing and reconstructing targets through a generated built upon U-Net. The method significantly improves intersection over union values of the detection results compared to state-of-the-art methods, showing outstanding performance on various backgrounds and targets.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Engineering, Multidisciplinary
Luyao Wang, Xuewen Wang, Bo Li
Summary: Recent advancements in the coal industry have led to the development of intelligent visual coal-gangue sorting, which combines deep learning and machine vision to improve the accuracy of coal-gangue detection. In this paper, a coal-gangue detection model driven by data optimization for multi-target detection tasks is proposed, along with a multi-object coal-gangue image synthesis model called BSP. The proposed model improves detection accuracy by 7% compared to the original SSD model and achieves almost two times faster detection with higher accuracy compared to the structure-optimized object detection model.
