Article
Computer Science, Information Systems
Cheol-Ho Choi, Hyun Woo Oh, Joonhwan Han, Jungho Shin
Summary: This study introduces a cell-based disparity refinement processor that extends a previous HMMF processor. Experimental results demonstrate that our proposed processor outperforms conventional processors on the KITTI 2012 and 2015 stereo benchmark datasets. Furthermore, our processor exhibits superior refinement performance when applied to the Cityscapes and StereoDriving datasets. Additionally, in terms of hardware resource utilization, our proposed processor shows lower resource requirements than conventional processors when implemented on an FPGA.
Article
Computer Science, Information Systems
Peng Yao, Jieqing Feng
Summary: This paper aims to reverse the inferiority of conventional algorithms in stereo matching by leveraging Stacking Learning with Coalesced Cost Filtering and demonstrating superior performance compared to challenging stereo matching algorithms on Middlebury v.2 and v.3 datasets. The proposed algorithm even outperforms deep learning methods in online results.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2021)
Review
Engineering, Civil
Aditya Pandey, Ashmeet Singh, Paolo Gardoni
Summary: This paper reviews the diagrammatic perturbation theory, a technique in Information Field Theory, for analytically estimating moments of perturbative non-Gaussian distributions. When dealing with physical phenomena, which often exhibit non-Gaussian features, approximation of the underlying distribution and inference of its parameter form are commonly used. More rigorous analysis methods such as Markov Chain Monte Carlo can also be employed, but are computationally expensive.
Article
Computer Science, Information Systems
Hui-Yu Huang, Zhe-Hao Liu
Summary: Stereo matching is a challenging problem in computer vision, especially in areas like 3DTV and 3D visualization. A proposed spatiotemporal disparity refinement method effectively addresses flickering errors caused by estimated disparity sequences, improving video quality and achieving high peak signal-to-noise ratio compared to state-of-the-art methods.
Article
Optics
Tao Yang, Rui Yang, Yuehong Qiu
Summary: This paper introduces a novel multi-brightness layer mechanism with a genetic optimization algorithm to address the challenge of stereo matching under dramatic illumination changes. The mechanism effectively reduces illumination variations and improves efficiency and accuracy in stereo matching. Additionally, an enhanced genetic optimization algorithm is designed to further enhance accuracy and stability in stereo matching, leading to better performance compared to state-of-the-art methods.
Article
Computer Science, Hardware & Architecture
Li Dong, Yong Han, Maohai Hu, Hao Luo, Yi Wang
Summary: This paper proposes a new high-precision stereo matching method, which achieves accurate occlusion recovery and discontinuity preservation through disparity refinement and techniques such as adaptability-based disparity reconstruction and disparity-based secondary guided filtering. Experimental results show that this method performs well in terms of matching accuracy and is less sensitive to the accuracy of disparity estimation.
Article
Engineering, Multidisciplinary
Mingyue Liu, Huizhong Zhu, Xinchao Xu, Youqing Ma, Shuo Zhang, Junbiao Wang
Summary: This paper presents a fusion method of stereo vision and IMU for precise measurement of the velocity, angle, and angular velocity of a moving carrier. The method achieves high measurement accuracy through self-calibration and three-dimensional reconstruction. It can provide reliable technical support for estimating the motion state of deep space moving vehicles.
Article
Chemistry, Analytical
Victor H. H. Diaz-Ramirez, Martin Gonzalez-Ruiz, Vitaly Kober, Rigoberto Juarez-Salazar
Summary: A stereo matching method based on adaptive morphological correlation is proposed. The method determines the point correspondences of an input pair of stereo images by matching locally adaptive image windows using an optimal morphological correlation. The proposed method accurately determines the point correspondences in homogeneous image regions and at the edges of scene objects. It is also capable of recovering unknown correspondences of occluded and not matched points in the scene using a simple post-processing.
Article
Engineering, Electrical & Electronic
Benyamin Kheradvar, Amir Mousavinia, Amir M. Sodagar
Summary: This paper proposes a new ground plane detection method based on the concept of iso-disparity strips and uses Cumulative Moving Average and dynamic thresholding techniques to enhance robustness. Experimental results show that the method can detect the ground plane within 65 milliseconds and 224 milliseconds, with low false positive rates and high true positive rates.
IEEE SENSORS JOURNAL
(2021)
Article
Chemistry, Analytical
Ming Wei, Ming Zhu, Yi Wu, Jiaqi Sun, Jiarong Wang, Changji Liu
Summary: The paper introduces a new end-to-end fast deep learning stereo matching network that innovates in feature extraction and cost volume construction, while combining edge guidance and multi-cross attention model to achieve excellent performance in both speed and accuracy.
Article
Biology
Ignacio Quintero, Marc A. Suchard, Walter Jetz
Summary: This study characterizes the evolution of bird species' temperature and precipitation niche spaces and finds that extant birds evolved from warm, mesic climatic niches to colder and drier environments. The overall rates of niche evolution have steadily increased, with some lineages experiencing exceptionally high rates due to the colonization of new niche spaces. The findings highlight the importance of integrating comprehensive environmental and phylogenetic information in ecological and conservation studies.
PROCEEDINGS OF THE ROYAL SOCIETY B-BIOLOGICAL SCIENCES
(2022)
Article
Engineering, Electrical & Electronic
Weimin Yuan, Cai Meng, Xiaoyan Tong, Zhaoxi Li
Summary: F-GDGIF effectively alleviates halo artifacts by incorporating efficient multi-scale edge-aware weighting and reduces computational costs by sub-sampling strategy, resulting in more accurate disparity maps with lower computational cost in stereo matching.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2021)
Article
Computer Science, Information Systems
Sungan Yoon, Jeongho Cho
Summary: This study proposes a method to improve the learning performance of a CNN-based object detection model by generating synthetic data and enhancing the learning data. Additionally, the model's detection performance is further improved through the fusion of stereo-vision cameras' transformed disparity map.
CMC-COMPUTERS MATERIALS & CONTINUA
(2022)
Article
Computer Science, Artificial Intelligence
Xiaowei Yang, Zhiguo Feng, Yong Zhao, Guiying Zhang, Lin He
Summary: The research proposes RDNet, which incorporates edge cues into stereo matching and generates a depth ground-truth boundary dataset by mining instance segmentation and semantic segmentation datasets. Additionally, methods like multi-scale cost volume and disparity refinement network are introduced to further optimize stereo matching performance.
IMAGE AND VISION COMPUTING
(2022)
Article
Computer Science, Information Systems
Jie Wang, Chenglei Peng, Ming Li, Yang Li, Sidan Du
Summary: This research focuses on accelerating stereo matching algorithms and improving the accuracy of disparity estimation using a multi-baseline trinocular camera model. Special schemes were designed to narrow the disparity searching range and optimize the steps of matching cost calculation, cost aggregation, and disparity computation. The evaluation results showed that the proposed schemes significantly reduced computational complexity and improved accuracy.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Jiaqi Dong, Zeyang Xia, Weiwu Yan, Qunfei Zhao
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2019)
Article
Computer Science, Artificial Intelligence
Ye Huang, Chaochen Gu, Xinping Guan
Summary: This paper proposes a Proportional-Integral (PI) neural network architecture that combines integral control and linear control, further improving the sample efficiency and training performance on most RL tasks. Experimental results demonstrate that the proposed architecture outperforms generally used MLP and other existing applied models on public RL simulation platforms.
NEURAL PROCESSING LETTERS
(2021)
Article
Computer Science, Artificial Intelligence
Changsheng Lu, Chaochen Gu, Kaijie Wu, Siyu Xia, Haotian Wang, Xinping Guan
Article
Computer Science, Artificial Intelligence
Bao-Qing Yang, Xin-Ping Guan, Jun-Wu Zhu, Chao-Chen Gu, Kai-Jie Wu, Jia-Jie Xu
Summary: The paper presents a discriminative dictionary learning framework based on support vector machines and feedback mechanism to enhance image classification performance.
PATTERN RECOGNITION
(2021)
Article
Automation & Control Systems
Jing Xiong, Changfu Xu, Khalil Ibrahim, Hao Deng, Zeyang Xia
Summary: This study introduces a mechanism-image fusion approach for calibrating a US-guided dual-arm robotic brachytherapy system, eliminating the need for external trackers and complex phantoms. The calibration accuracy of the system is 0.65 +/- 0.31 mm, outperforming state-of-the-art methods according to experimental results and comparisons.
IEEE-ASME TRANSACTIONS ON MECHATRONICS
(2021)
Article
Chemistry, Multidisciplinary
Jiaqi Dong, Zeyang Xia, Qunfei Zhao
Summary: The paper proposes an algorithm for dynamic gesture recognition and prediction in augmented reality assisted assembly training, aiming to evaluate the standard and achievement of hand operations. By decomposing tasks into hand operations and further into continuous actions, each related to a standard gesture, the algorithm improves recognition accuracy.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Artificial Intelligence
Tingman Yan, Xiaolin Huang, Qunfei Zhao
Summary: This paper proposes a hierarchical superpixel segmentation method based on the 1-nearest neighbor (1-NN) graph of pixels/superpixels. The method ensures connectivity by building 1-NN graphs from pixel/superpixel adjacent matrices. The weakly connected components (WCCs) of the 1-NN graph are labeled as superpixels to determine the next-level hierarchy. The paper also introduces a two-stage parallel labeling method based on the forest-like structure of the WCCs. Experimental results show that the proposed method has comparable performance and is several times faster than other algorithms.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Fei Zhang, Chaochen Gu, Chenyue Zhang, Yuchao Dai
Summary: This paper introduces a novel Complementary Patch representation based on information theory, which generates Class Activation Maps with more information related to object seeds by using a pair of input images with complementary hidden parts. By constructing a CP Network and a Pixel-Region Correlation Module, the quality of CAM segmentation can be further improved.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Automation & Control Systems
Yuwei Xu, Lijuan Feng, Zeyang Xia, Jing Xiong
Summary: This paper proposes an end-to-end CNN-based network to deal with the challenges in detecting and describing feature keypoints in robotic gastrointestinal endoscopy, demonstrating effective results in challenging conditions.
INTELLIGENT ROBOTICS AND APPLICATIONS, ICIRA 2021, PT III
(2021)
Proceedings Paper
Automation & Control Systems
Huixian Peng, Lei Deng, Zeyang Xia, Yaoqin Xie, Jing Xiong
Summary: A method for respiratory motion characterization based on RGB-D camera and B-spline elastic registration is proposed for accurate treatment planning in robotic radiosurgery dealing with thoracic or abdominal tumors. It uses depth camera images to model abdomen surface during respiration and applies B-spline elastic registration technique to restrict the measuring area anatomically. The feasibility of the method and device is verified through error analysis and shape comparison.
INTELLIGENT ROBOTICS AND APPLICATIONS, ICIRA 2021, PT III
(2021)
Proceedings Paper
Automation & Control Systems
Hao Cheng, Chaochen Gu, Kaijie Wu
2020 4TH INTERNATIONAL CONFERENCE ON CONTROL ENGINEERING AND ARTIFICIAL INTELLIGENCE (CCEAI 2020)
(2020)
Article
Automation & Control Systems
Changjian Gu, Chaochen Gu, Kaijie Wu, Liangjun Zhang, Xinping Guan
INTERNATIONAL JOURNAL OF ROBOTICS & AUTOMATION
(2019)
Proceedings Paper
Computer Science, Artificial Intelligence
Shuxin Zhao, Chaochen Gu, Changsheng Lu, Ye Huang, Kaijie Wu, Xinping Guan
2019 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN)
(2019)
Article
Computer Science, Information Systems
Yangzhou Gan, Jing Xiong, Qunfei Zhao, Zeyang Xia