作者： Lingfei Song;Hua Huang （¹School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China ²Haihe Laboratory of ITAI, Tianjin, China ³School of Artificial Intelligence, Beijing Normal University, Beijing, China）

出处： IEEE Transactions on Image Processing 2023 Vol.32 P1065-1077

关键词： Noise reduction;Maximum likelihood estimation;Image denoising;Approximation algorithms;Signal processing algorithms;Task analysis;Noise measurement

摘要： Digital images often suffer from the common problem of stripe noise due to the inconsistent bias of each column. The existence of the stripe poses muc ...

作者： Lingfei Song;Hua Huang （¹School of Computer Science and Technology, Beijing Institute of Technology, China ²Haihe Lab of ITAI ³School of Artificial Intelligence, Beijing Normal University, China）

出处： IEEE transactions on pattern analysis and machine intelligence 2023 P1-14

关键词： Dark current;Signal to noise ratio;Entropy;Temperature measurement;Maximum likelihood estimation;Sensors;Minimization

摘要： Due to the manufacturing imperfections, nonuniformities are ubiquitous in digital sensors, causing the notorious Fixed Pattern Noise (FPN). The abilit ...

作者： Yaohui Zhu;Xiaoyu Sun;Miao Wang;Hua Huang （¹School of Artificial Intelligence, Beijing Normal University, Beijing, China ²School of Computer, Beijing Institute of Technology, Beijing, China）

出处： IEEE Transactions on Intelligent Transportation Systems 2023 P1-12

关键词： Transformers;Feature extraction;Object detection;Fuses;Semantics;Visualization;Standards

摘要： RGB-Infrared multi-modal object detection utilizes diverse and complementary information, showing some advantages in intelligent transportation field. ...

发明人： 张磊，董彪，赵天琦，黄华

申请人： 北京理工大学

申请号： 202210323200.5

申请日期： 2022.03.29

摘要： 本发明涉及一种抑制语音要素异常点的文本驱动语音合成的方法，属于语音信号处理和人工智能的技术领域。首先，以一种更具鲁棒性的注意力对齐机制，实现音素到梅尔频谱图的对齐，在音素长度扩展到梅尔频谱图长度的过程中，利用截断误差计算，能够有效避免极端值对整体数据的影响，使数据的描述结果更加合理与稳定。然后，采用 ...

发明人： 张磊，孙心桐，黄华

申请人： 北京理工大学

申请号： 202210317968.1

申请日期： 2022.03.29

摘要： 本发明涉及一种基于灰点漂移的视频白平衡方法，属于图像处理中的颜色恒常性与白平衡技术领域。本方法在光源估计的过程中，除图像本身的单帧光源估计，同时考虑了其相邻帧图像的光源估计、稳定视频的颜色。通过加权融合的方式调整已有估计和单帧估计的比例，使视频白平衡结果兼顾正确性与稳定性。在融合权重的确定时，除了考 ...

发明人： 张磊，董彪，黄华

申请人： 北京理工大学

申请号： 202211291702.0

申请日期： 2022.10.20

摘要： 本发明涉及一种基于特征金字塔的文本驱动语音合成方法，属于语音信号处理和人工智能技术领域。本方法从音频频谱图中提取能量和音高的特征信息，分别以均方根能量和基音频率进行提取，对应于响度和音调的声音元素，作为底层特征。同时，从通过继承音色的声音元素的梅尔谱图的时频分析得到时频信息，分别以过零率与谱质心进行 ...

作者： Feng, Hansen¹; Wang, Lizhi¹; Wang, Yuzhi²; Huang, Hua¹

出处： 30th ACM International Conference on Multimedia, MM 2022 Lisboa, Portugal 2022

会议录： 1436-1444

作者： Song, Lingfei;Huang, Hua （¹School of Computer Science and Technology, Beijing Institute of Technology, Beijing; 100081, China）

出处： arXiv 2022

摘要： Competitive Coding approach (CompCode) is one of the most promising methods for palmprint recognition. Due to its high performance and simple formulat ...

作者： Wei, Kaixuan;Aviles-Rivero, Angelica I.;Liang, Jingwei;Fu, Ying;Huang, Hua;Sch?nlieb\', Carola Bibiane （¹School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China;²Department of Pure Mathematics and Mathematical Statistics, University of Cambridge, Cambridge, United Kingdom;³Institute of Natural Sciences and School of Mathematical Sciences, Shanghai Jiao Tong University, Shanghai, China;⁴School of Artificial Intelligence, Beijing Normal University, Beijing, China;⁵Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Cambridge, United Kingdom）

出处： Journal of Machine Learning Research 2022 Vol.23

摘要： Plug-and-Play (PnP) is a non-convex optimization framework that combines proximal algorithms, for example, the alternating direction method of multipl ...

作者： Xu, Yanchao¹; Shao, Wenbo²; Li, Jun²; Yang, Kai³; Wang, Weida¹; Huang, Hua¹; Lv, Chen⁴; Wang, Hong²

出处： 25th IEEE International Conference on Intelligent Transportation Systems, ITSC 2022 Macau, China 2022

会议录： Vol.2022-October 2471-2478