电子工程系

Department of Electronic Engineering


马惠敏,女,清华大学电子工程系副教授,博士生导师,三维图像仿真研究室(3D Image Lab)负责人,中国图象图形学学会秘书长,北京市“三八红旗奖章”获得者。

实验室主页:http://3dimage.ee.tsinghua.edu.cn
Email:mhmpub@tsinghua.edu.cn 
电话:010-62781432


 

个人简介:

研究方向为三维图像认知,将认知心理学稳定的高层语义先验引入机器学习,研究复杂环境中的目标图像检测、认知、导航等科学问题。

作为负责人承担了国家自然科学基金项目,专项重点基金项目,国家重点研发计划子课题,国际国内企业合作项目20余项。近年作为通讯作者发表机器视觉与模式识别领域高水平SCI期刊(TPAMI、PR等)和顶级国际会议(CVPR等)论文十余篇,并实现了从基础理论、核心技术到产业的全链条的突破性进展。

2015年,在国际计算机视觉算法评测数据集(KITTI)目标检测国际评测中全部评测六项指标中,四项获得第一,两项获得第三;2016年,基于图像认知的心理测评方法及系统获得吴文俊人工智能科学技术创新奖一等奖;2017年图像认知心理系统获日内瓦国际发明展银奖。

 

教育背景

2001年在北京理工大学机电工程学院获得博士学位;
2011年在美国匹兹堡大学医学院做访问学者。

 

工作履历

(1) 2001-2006年,清华大学电子工程系助理教授
(2) 2006年至今,清华大学电子工程系副教授

教学经历:
“数字逻辑与处理器基础”,本科生核心课,2012年至今
“数字图像技术及应用”,研究生课,2002年至今
“微机原理”,本科生基础课,2007-2011年


学术兼职

Organizing Chair, the 9th International Conference on Image and Graphics, 2017
Organizing Chair, the 4th International Symposium on Bioelectronics and Bioinformatics, 2015
Organizing Committee Chair, Image and Graphics Technology and Application, 2013, 2014, 2015
Conference Chair, IASTED International Conference on Modelling, Simulation and Identification, 2009
Program Chair , IASTED A sia Conference on Modelling and Simulation, 2007


社会兼职

2016年至今,  中国图象图形学学会秘书长、常务理事
2016年至今,  北京图象图形学学会监事长
2012年至今,  北京市科协委员
2011年至今,  IEEE会员
2007-2016年,北京图象图形学学会秘书长、常务理事
2004-2016年,中国图象图形学学会青年工作委员会主任


研究领域

三维图像认知理论:将高层认知心理学理论与图像识别相结合,以多任务、多模态、多视角认知模型为核心,解决强遮挡、多尺度、目标姿态变化大等三维物体检测与识别中的难题。
视觉导航技术:结合导航任务对复杂环境进行仿真建模,以三维图像认知理论及最优化理论为核心,实现复杂环境下的目标识别、定位及导航决策。
应用领域:智能无人系统


研究概况

(1) 三维图像认知理论

针对数据驱动的机器学习在不确定的复杂场景图像识别中的困难,从格式塔认知心理学寻找突破,通过对人的试验发现稳定的认知主体语义先验,建立认知先验数学模型,创新性地在基于自底向上的数据驱动的机器学习中引入自顶向下的主体先验的驱动,提出了“数据和主体双向驱动认知学习”的学术思想,强调人作为主体的先验知识在机器学习中的诱导作用,建立了三维图像格式塔认知理论,显著提高了机器学习对复杂场景的识别能力。

 

 

(2) 基于图像认知的心理测评及干预系统

首次在国际上提出了心理学语义到图像的映射,建立了第一个心理图像库,提出了基于图像认知的心理特征量化提取方法,改变了传统的问题式的主观测评模式,实现了基于视觉的客观、准确、快速的心理测评模式,可用于心理问题的早期发现以及视觉注意机制的理论研究。教育部鉴定为原始性创新,国际领先水平,是社会重大需求。“基于图像认知的心理测评方法及系统”2016年获得吴文俊人工智能科学技术创新奖一等奖,2017年获得日内瓦国际发明展览会银奖。

 

基于图像认知的心理测评及干预系统


(3) 三维图像认知与视觉导航技术

提出了3D图像认知计算理论模型,定义了结构、环境先验等通用认知特性,将多任务、多模态、多视角3D环境主体先验引入机器学习,实现了数据与主体双向驱动认知学习,用于复杂驾驶场景3D物体检测与定位,2015年在国际最大的自动驾驶权威评测数据集KITTI上,共六项获得四项第一,首次将物体检测准确率从80%提高到90%。

 

 自动驾驶KITTI测评三维物体检测


(4) 三维复杂环境仿真及导航决策应用

通过建立复杂环境及干扰遮挡的三维模型和物理模型,构建了遮挡图像库及导航任务动态图像序列库,建立了高动态复杂环境仿真系统,为视觉导航提供感知学习样本、认知学习平台、增强学习环境、智能决策可视化等,以实现在干扰遮挡等情况下的三维目标检测识别及智能导航决策,作为智能无人系统视觉导航的核心技术,在航空、航天、船舶领域实际应用中发挥了重要作用。

 

 复杂环境仿真及增强学习架构


主要研究项目
      基于机器视觉的环境目标检测与识别与驾乘状态监测、评估和意图识别,国家重点研发计划
      适应复杂环境的红外成像导航技术研究,专项重点基金
      基于GPU高动态复杂环境仿真,专项重点基金
      基于图像认知的人的心理特征的提取与分析,国家自然科学基金
      用于三维目标识别的形态图理论和方法研究,国家自然科学基金
      三维系统仿真及图像识别与跟踪系统设计,国家专项系列项目
      CNN for bus/LRT Management 国际合作项目
      辅助/自动驾驶视觉信息处理系统,企业合作项目

 

奖励与荣誉

2017年,获得日内瓦国际发明展览会银奖,排名第一
2016年,获得吴文俊人工智能科学技术创新奖一等奖,排名第一
2015年,自动驾驶KITTI国际评测中目标检测获四项第一,两项第三


学术成果

期刊文章

1. Xiaozhi Chen, Kaustav Kundu, Yukun Zhu, Andrew Berneshawi, Huimin Ma, Sanja Fidler and Raquel Urtasun, 3D Object Proposals for Accurate Object Class Detection, IEEE Transactions on Pattern Analasys and Machine Intelligence (PAMI), 2017
2. Xi Li, Huimin Ma, Traffic Light Recognition for Complex Scene with Fusion Detection, IEEE Transactions on Intelligent Transportation Systems (ITS), 2017
3. Xiaozhi Chen, Huimin Ma, Chenzhuo Zhu, Xiang Wang, Zhichen Zhao, Boundary-Aware Box Refinement for Object Proposal Generation, Neurocomputing, 2016
4. Zhichen Zhao, Huimin Ma,Xiaozhi Chen, Generalized Symmetric Pair Model for Image Recognition, Pattern Recognition (PR), 2016 
5. Zhichen Zhao, Huimin Ma,Xiaozhi Chen, Semantic parts based top-down pyramid for action recognition, Pattern Recognition Letter (PRL), 2016
6. Xiang Wang, Huimin Ma,Xiaozhi Chen, Geodesic Weighted Bayesian Model for Saliency Optimization, Pattern Recognition Letter (PRL), 2016
7. Jinghuan Wen, Huimin Ma, Xiaoqing Zhang, Optimization of the Occlusion Strategy in Visual Tracking, Tsinghua Science and Technology, 2016
8. You Shaodi, Ma Huimin, Manifold topological multi-resolution analysis method, Pattern Recognition (PR),2011
9. Huimin Ma, Tiantian Huang, Yanzhi Wang, Multi-resolution recognition of 3D object based on visual resolution limit, Pattern Recognition Letter (PRL), 2010
10. Guo Yuchen, Ma Huimin, Liao Qingmin, Analysis on Shape Features for 2D Projective Image Based 3D Object Recognition, Journal of Tsinghua University, 2009
11. Huang Wenyun, Ma Huimin, Wang Shengjin, A Recognition Algorithm for Sea Background Infrared Target, Journal of Tsinghua University (SCi & Tech ), 2009
12. Yu Xiaoliang, MA Huimin, XIAO Jian, Real-time Target Recognition System Simulation Based on Laser Near-field Detection, Optoelectronics Letters, 2009
13. Wu Chenye, Ma Huimin, Computing EEE Events Based on Conicoid Theory in 3D Object Recognition Application, Tsinghua Science and Technology, 2009
14. Sun Jie, Ma Huimin, Li Fengting, "Viewpoint Space Partition Based on Affine Invariant Feature," Journal of Tsinghua University (SCi & Tech ), 2009
15. Su Miao, Ma Huimin, Li Fengting, Catastrophe Theory and Clustering Algorithm Based Viewpoint Space Partition Algorithm, Journal of Tsinghua University, 2008
16. Ma Huimin, Huang Ruke, Multi-hierarchical Algorithms for Fast Fingerprint Identification, Chinese Journal of Eelectronics, 2005.
17. Wang Huajian, Zhang Aiwu, Ma Huimin, Li Fengting, Hole Filling in Range Image Processing Based on Statistical Model, Journal of Optoelectronics ?Laser, 2004
18. Ma Huimin, Wang Yan, Parallel Feature Extraction System, Chinese Journal of Electronics, 2003
19. Ma Huimin,Su Guangda,High Speed Recognition System of Plane Infrared Image, Journal of Infrared and Millimeter Waves, 2002
20. Ma Huimin, Gray Image Feature Extraction Base on Projection, Optical Technique, 2002

会议文章

1. Xiaozhi Chen, Huimin Ma, Ji Wan, Bo li, Tian Xia, Multi-View 3D Object Detection Network for Autonomous Driving, CVPR: 30th IEEE Conference on Computer Vision and Pattern Recognition, Jun, 2017, Honolulu, USA
2. Zuoquan Li, Huimin Ma, Zhengyu Liu, Road Lane Detection With Gabor Filters, The International Conference on Computer Technology and Software Engineering (CTSE), May 13th - 15th, 2016, Wuhan, Hubei, China
3. Wang Xiang, Ma Huimin, Chen Xiaozhi,Salient Object Detection Via Fast R-CNN And Low-level Cues.International Conference on Image Processing (ICIP), September 25-28, 2016, USA
4. Zhao Zhichen, Ma Huimin, Chen Xiaozhi,Multi-scale Region Candidate Combination for Action Recognition. International Conference on Image Processing (ICIP), September 25-28, 2016,USA
5. Shi Dazhao, Ma Huimin, A Robust Object Tracking Algorithm in Complex Environment, 8th International Conference on Digital Image Processing (ICDIP), May 20-23,2016, Chengdu, China
6. Xiaozhi Chen, Kaustav Kundu , Huimin Ma, Sanja Fidler, Raquel Urtasun, Monocular 3D Object Detection for Autonomous Driving, CVPR: 29th IEEE Conference on Computer Vision and Pattern Recognition, Jun.26-30, 2016, Las vegas, USA
7. Yidong Wang, Huimin Ma, Identification differences among people under context of complex images, 4th Intenational Symposium on Bioelectronics and Bioinformatics (ISBB), October 14-17, 2015, Beijing, China.
8. Xiaozhi Chen, Yukun Zhu , Kaustav Kundu, Huimin Ma ,Sanja Fidler, Raquel Urtasun, 3D Object Proposals for Accurate Object Class Detection, The Twenty-ninth Annual Conference on Neural Information Processing Systems (NIPS), December 7-12, 2015, Montreal, Canada
9. Xiang Wang, Huimin Ma, Jiayun Hou, Viewpoints selection of 3D object recognition based on Manifold topological multi-resolution analysis method (IGTA), June.19-20, 2015, Beijing, China
10. Xiang Wang, Huimin Ma, Xiaozhi Chen, Geodesic weighted Bayesian model for salient Object Detection. International Conference on Image Processing (ICIP), Sep.27-30, 2015, Quebec City, Canada
11. Cao Jun, Ma Huimin, Regression-based Parameter Optimization for Binary Output Systems (ICoICT), May 27-29, 2015, Bali, Indonesia
12. Xiaozhi Chen, Huimin Ma,Improving Object Proposals with Multi-Thresholding Straddling Expansion, CVPR: 28th IEEE Conference on Computer Vision and Pattern Recognition, Jun.7-12, 2015, Boston, USA
13. Boqiang Fan, Huimin Ma, Xiang Wang, The Analysis of Frequency Domain Characteristics of Emotional Images in Eye-tracking Experiment, 7th International Conference on Digital Image Processing (ICDIP), April 9-10,2015, Los Angeles, USA
14. Chen Xiaozhi, Ma Huimin, Learning a Compact Latent Representation of the  Bag-of-parts Model, International Conference on Image Processing (ICIP), Oct. 25-31, 2014, Paris, France
15. Zhao Zhichen, Ma Huimin, Chen Xiaozhi, Protected Pooling Method Of Sparse Coding In Visual Classification. International Conference on Computer Vision and Graphics (ICCVG), September 15-17, 2014, Warsaw, Poland
16. Wenyu Li, Huimin Ma, Xiang Wang, Dazhao Shi, Features Derived From Behavioral Experiments To Distinguish Mental Healthy People From Depressed People, IASTED International Conference on Biomedical Engineering (ICBE), Jun 23-25, 2014, Zurich, Switzerland
17. Bao Shurui, Ma Huimin, Li Wenyu, Wang Xiang, Discrimination of Positive Facial Expressions is More Susceptible to Affective Background Scenes, 2014 International Conference on Culture, Knowledge and Society (ICCKS), January 24-25, 2014, Macau, China
18. Bao Shurui, Ma Huimin, Li Wenyu, ThuPIS: A New Affective Image System for Psychological Analysis, 2014 IEEE International Symposium on Bioelectronics and Bioinformatics (ISBB), April 11-14, 2014, Taiwan, China
19. Luyao Zhou, Huimin Ma,Accurate and Fast Simulation of Laser Scanning Imaging, 3rd International Conference on Multimedia Technology (ICMT), Nov.29 - Dec. 1, 2013, Guangzhou, China
20. Ma Huimin, Chen Xiaozhi, Simplification Computing for Visual Events Based on Distinct Silhouette and View-Independent Pruning Algorithm, Image and Graphic Technology and Application (IGTA), Volum 363, April 2-3 2013, Beijing, China
21. Zhangyu, Ma Huimin, The Electronic Countermeasures optimization based on PCA and multiple regression analysis, Image and Graphic Technology and Application (IGTA), Volum 363, April 2-3 2013, Beijing, China
22. Ouyang Shangfu, Ma Huimin, Cognitive impact differs from relationship of scene and facial expression The 2012 International Conference on Biomedical Engineering and Biotechnology (ICBEB), May 28th -30th, 2012, Macau, China
23. Li Juan, Ma Huimin; Zhou Luyao; Implementation of a Progressive Target Recognition System for All the Angles. The International Conference on Automatic Control and Artificial Intelligence, March 24-26, Xianmen, China
24. Hou Yu, Ma Huimin. Analysis of an infrared interference antagonizing window tracking algorithm. International Conference on Automatic Control and Artificial Intelligence, March 24-26, 2012, Xianmen, China
25. Guanhua Su, Huimin Ma, Yu Hou, A Robust Approach for Anti-Jamming Target Tracking in Forward Looking Infrared Imagery, Sixth International Conference on Image and Graphics (ICIG),12-15 Aug. 2011, China
26. Pang Bo, Ma Huimin, An Effective Way of 3D Model Representation in Recognition System, 2011 International Conference on Multimedia and Signal Processing (CMSP),2011.6 Guilin, China
27. Liu Lifeng, Ma Huimin, Lu Mingquan, A FPGA and Zernike Moments Based Near-field Laser Imaging Detector Multi-scale Real-time Target Recognition Algorithm, Third International Symposium on Information Science and Engineering (ISISE), 24-26, Dec.2010, Shanghai, China
28. Xiao Yu, Huimin Ma and Ze Yuan, A Solution to Efficient Viewpoint Space Partition in 3D Object Recognition, International Conference on Image and Graphics (ICIG), Xian, China,September 21-23, 2009.
29. Xu-Zhang, Hui-min Ma, Yu-shu Liu, Chun-xiao Gao,An Extension of Locally Linear Embedding for Pose Estimation of 3D Object, ICMLC, Aug. 19-22, 2007, Hongkong, China
30. Huang Tiantian, Ma Huimin, Li Fengting, Segmentation and Recognition of 3D Objects in Automatic Navigation, The 4rd International Conference on Natural Computation (ICNC), Oct.18, 2008, Jinan China
31. Wu Chenye, Ma Huimin, Efficiently Pruning for Visual Events in 3D Object Recognition Application, 14th National Conference of Image and Graphics, Fuzhou, China, May 16-18, 2008, pp259-264.
32. Zhang Yibo, Ma Huimin, Cui Peng, Manifold Characteristic Analysis of 3-D Object,  2008 Conference of Image and Graphics (NCIG), Beijing, April 20, 2008.
33. Yuan Luo, Huimin Ma, Fengting Li, An Improved Algorithm of Viewpoint Space Partition in 3D Object Recognition Application, the Seventh IASTED International Conference on Signal and Image Processing (SIP), Aug. 14-16, Honolulu, Hawaii, USA
34. Y. Luo, H. Ma, F. Li, 3D Object Recognition Technique Based on Muti-Resolution Aspect Graph, 2005 International Conference on Neural Networks and Brain (ICNNB), October 13-15 Beijing China
35.  W. Lin, H. Ma, F. Li, Accurate Tracking Based on Interest Points in The Infrared Imaging Guidance, International Conference on Neural Networks and Brain (ICNNB), October 13-15, 2005, Beijing China
36. Ma Huimin, Huang Ruke, A New Fingerprint Translation Finding Algorithm, IEEE 2004 International Conference on Machine Learning and Cybernetics (ICMLC), Shanghai, China
37. Ma Huimin, Su Guangda, Huang Ruke, A Classification Agorithm for Fingerprint Matching in Large Database, The 13th National Conference on Neural Network (NCNN), November 12-15, 2003, Qingdao, China
38. Ma Huimin, Su Guangda, Wang Junyan, A Glass Bottle Defect Detection System Without Touching, IEEE, 2002 International Conference on Machine Learning and Cybernetics (ICMLC), , November 4-5, 2002,Beijing, China

1) 《数字图像处理》,高等教育出版社,2016,副主编,并撰写第八章。
2) 《数字图像处理》,北京大学出版社,2007,副主编,并撰写第八章
3) 《视觉盛宴》,北京出版社,2015,撰写第一章