Current Location:homepage  Research
Pattern Recognition and Intelligence System Research Center
Editor:贾岩  Updated:2016-01-07  Views:187

1. Mission

1) Man-machine symbiosis in uncertain environments

In man-machine game field, lettinga machine-player learn to play curling is more difficult than to play chess, as AlphaGo series.

As one may know, the environment of ice curling track is with high uncertainty, i.e. dynamic space and time changing of local temperature and frictional resistance.Curling strikingneed to solve problems of high dimensionalprediction and game strategy under the continuous physical space, continuous state space, and continuous action space as well.“Curling Champion” is an intelligent curling computing platform designed for the study and validation of intelligent prediction and decision algorithms. Basically, reinforcement learning, CNNs, and other variations of classical algorithms could be joined in the platform, which is stillin its perfection, for training, testing and validating.

2)  Intelligent vision system and active cybernation

Deep space exploration mission oriented small body measurement and modeling,vision based navigation and on-board intelligentservices arethe recent and future imperative tasks of China Aerospace, among which several key scientific research been developed  or on the road to be conquered.

Flying by the near-earth asteroid, called Toutatis, no. 4179, is the second extended mission of Chinese famous deep space prober Chang 'e 2, launched in 2010. During the fly-by, valuable information of this small body was captured as image sequence, with which relevant vision measurement, 3D data fusion and remodeling and small body surface science research are developed. Research outputs could be seen in papers published in Nature - Scientific report, Planetary Space Science etc. journals and workshops. Recently, a specific images-based multi-resolution three dimensional measuring and relative navigation simulating platform is building and testing for future deep space exploration of China. This is the very first time Chinese scientists and engineers going in for the challenging mission of visiting, landing on and sample returning from a solar system small body.

2. Overview

The research field of pattern recognition is a study of how machines observe the environment, how they distinguish the patterns of interest, and how they make sound and reasonable decisions about the categories of the patterns.

The research in this field involves the broad field of pattern recognition, with specific interests in optical character recognition, active vision, biometric recognition, medical image processing, offline writer identification and much more. The faculty members of the school working in this field publish on a wide range of leading international and domestic journals and conferences such as IEEE Transactions on Image Processing, IEEE Transactions on Information Forensics and Security, Pattern Recognition, Pattern Recognition Letters, ICCV, CVPR, and ECCV. They are also active in program committees and editorial boards of these journals and conferences.

There are 6 professors, 10 associate professors and 5 lecturers working in this field, where 6 faculties are doctoral supervisors and 9 are master supervisors. Currently, there are more than 22 doctoral students and 58 master students studying in this field. Leading professors are: Prof. Xianglong Tang, Prof. Kuanquan Wang, Prof. Xiangqian Wu and Prof. Wangmeng Zuo.

In the past five years, 2011-2015, the faculty members in this field have been granted 20 research projects, among which 8 are funded by National Natural Science Foundation of China (NSFC), 1 funded by National High-tech R&D Program of China (863 Program). Also they have won 1 First Prize, 2 Second Prizes, 3 Third Prizes of Science-Technology Progress of Ministerial or Provincial Level. Over 600 research papers have been published in international and domestic academic journals and conferences, including over 100 papers published in  top journals and conferences.

100 Ph.D. students and more than 300 master studentsof this research field have graduated from the school. The outstanding alumni include Dr. Lei Xu, a fellow of European Academy of Sciences; Dr. Jinpeng Huai, a vice-minister of the Ministry of Industry and Information Technology (MIIT), and an academician of Chinese Academy of Sciences.

The detailed information of the research work in this field can be found in and

3. Research Topics

  • Active vision: Coping with problems such as occlusions, limited field of view and limited resolution of the camera.

  • Computer vision and image processing: Image denoising, visual saliency detection, object detection and recognition, sparse representation and metric learning, scene text detection and recognition, restoration and 3D reconstruction.

  • Biometric recognition: Palmprint, iris, face, signature and handwriting.

  • Medical image analysis: Retina image, OCT, ultrasonic image and mammography images.

4. The Faculty

Prof. Xianglong Tang

He is a professor and doctoral supervisor of the School of Computer Science and Technology, HIT. He received his B.S. and M.S. degrees from HIT in 1982 and 1986, respectively. He received his Ph.D. degree in computer science from HIT in 1995. Now he is the director of Pattern Recognition and Intelligence System Research Center at HIT and the academic leader of artificial intelligence and information processing. Currently, he serves as a board member of applied informatics. He is a member of Steering Committee Member of IScIDE (International Conference of Intelligence Science and Intelligent Data Engineering), a member of Chinese Association for Artificial Intelligence, vice president of Heilongjiang Association for Artificial Intelligence, a senior member of CCF (China Computer Federation), a member of Artificial Intelligence and Pattern Recognition Professional Committee of CCF and the director of Heilongjiang Province Computer Society Professional Committee of Human-computer Interaction.

He has wide and various research interests, but focusing on pattern recognition, machine learning, visual analysis of spacecraft equipment and medical image processing. He  has also offered the related  courses such as advanced pattern recognition and motion analysis and understanding for Ph.D students. He has implemented more than 20 projects from National Hi-tech R&D Program of China (863 Program), National Natural Science Foundation of China (NSFC), and some of international cooperation projects. He is the winner of 1 National Science and Technology Progress Award, 3 First Prizes, 1 Second Prize and 2 Third Prizes of Ministry Science and Technology Advancement Award. He has published more than 100 papers,1 monograph and 1 book, and has been granted 1 invention patent.

5 .Selected Publications

The faculty members of this field in the school publish their innovative findings, on top journals such as Transactions on Image Processing, IEEE Transactions on Information Forensics and Security, Pattern Recognition, Pattern Recognition Letters, and top conferences such as IJCV, IEEE T-IP, T-NNLS, T-IFS, CVPR, ICCV, ECCV, and NIPS.

  1. Zhixun Li, Yingtao Zhang, Guangzhong Liu, Haoyang Shao, Weimin Li, Xianglong Tang. A robust coronary artery identification and centerline extraction method in angiographies. Biomedical Signal Processing and Control, 2015, 16:1-8.

  2. Liming Yuan, Jiafeng Liu, Xianglong Tang. Combining Example Selection with Instanceselection to Speed up Multiple-instance Learning. Neurocomputing, 2014, 129(5):504-515.

  3. Yuru Wang, Xianglong Tang, Qing Cui. Dynamic Appearance Model for Particle Filter Based Visual Tracking. Pattern Recognition, 2012, 45(12): 4510-4523.

4. Yingtao Zhang, H. D. Cheng, Jianhua Huang, Xianglong Tang. An Effective and Objective Criterion for Evaluating the Performance of Denoising Filters. Pattern Recognition. 2012, 45(7): 2743-2757.

6 .Selected Research Projects

  1. National Natural Science Foundation of China (NSFC), Weakly Supervised Learning and its Applications on Online Learning and Objects Tracking (Grant No. 61173087). $0.104M, 2012.1-2015.12.

    This project focuses on a multi-view of weakly supervised learning based on a probability graph model to resolve a visual tracking online, improving the accuracy of recognition on dynamic objects. The project proposes a unified online learning framework to combine diverse weakly supervised problems. The main achievements include weakly supervised information measurement, the framework construction of a multi-view learning, an inference model based on probability graph, online learning and so on.

  2. National Natural Science Foundation of China (NSFC), Research on Palmprint Recognition in Complex Environments(Grant No. 61350004). $0.032M, 2014.1-2014.12.

    Computer-aided personal recognition becomes increasingly important in the information-based society, and, within this field, biometrics is one of the most important and reliable methods. The most widely used biometric feature is the fingerprint and the most reliable feature is the use of iris. However, it is very difficult to extract small unique features (known as minutiae) from unclear fingerprints, and the iris input devices are very expensive or incontinent. In contrast to all of these, the palm print seesseveral advantages. By combining all of the features of palms, such as palm geometry, ridge and valley features, and principal lines and wrinkles, it is possible to build a highly accurate biometrics system.

  3. National Natural Science Foundation of China (NSFC), The Computing and Visualization Method of Large-scale and Multimodal Virtual Heart (Grant No.61173086). 2012-2015.

    The visualization of the cardiac anatomy is essential to real-time diagnosis, medical training, surgical guidance and surgical tool tracking. In this project, an illustrative visualization method is presented via a perception based lighting enhancement. The proposed method can emphasize the given tissue while deemphasizing the other tissues in the human cardiac anatomy, and can provide the necessary user interactions to generate satisfied rendering results. Furthermore, the proposed method is integrated into a cardiac visualization system which exploits the parallelism of modern graphics processing units (GPUs).

  4. The Program of Ministry of Education for New Century Excellent Talents, Image Processing and Pattern Recognition. $0.08M, 2013.1-2015.12.

    This project focuses on, from a perspective of optimization and learning, the development of restoration models for recovering high-quality images from the problems of image blind deconvolution, non-uniform image deblurring, image super-resolution. It is geared to the development of more efficient and robust algorithms for recognition and classification of low-quality face and palmprint images.

  5. National Natural Science Foundation of China (NSFC), Research on Automated Diagnosis of Diabetic Retinopathy based on Image Recognition (Grant No.61073125). $0.054M, 2011.1-2013.12.

Currently, the scanning of retina is usually performed manually by ophthalmologist through a visual inspection of the retina, which is a time-consuming, repetitive, tiring, subjective, and error-prone process. Due to the limited available healthcare resources and the large number of potential patients (e.g. the diabetic patients and the old people), it is impossible for all these potential patients to be regularly checked manually by ophthalmologist, and hence many patients are exposed to more  vision impaired or even blind without a timely retina scanning and a proper treatment. To deal with this problem, this project investigatesan automated fashion of lesion detection and recognition in the retina images.

7 .Selected Awards

  1. Heilongjiang Province Natural Science Award, first-class prize. Analysis Techniques of Banknotes and its Application in the Series Products. Prof. Xianglong Tang. 2009.

    This award is for the work on techniques analyzing the banknotes and their  application to the series of products. For the first time, the multi-spectrum images of banknote can systematically be investigated. The algorithms of banknote image analysis are studied, including banknote classification, feature detection or verification (defect feature or anti-counterfeit feature), and quality evaluation, based on which, the system of banknote image processing has been achieved and it can be applied to many different environments.

  2. Science and Technology Development Ministry of Education of China, second-class prize.Breast Ultrasound based Computer-Aided Diagnosis system. Associated Prof. Yingtao Zhang. 2012.

    This award is for the research on a CAD system based on multi-mode breast ultrasound image, which provides exhaustively physiological and pathological information, increasing the accuracy of diagnosis. Not only it supports doctors with the objective and scientific diagnosis references, but also it relieves a physical and psychological pressure from patients.

8. Social Contribution

Banknote analysis technology is the core technology of modern financial optical and electrical equipment, such as automatic cash dispenser, banknote clearing machine and high-end banknote discriminator. This project combines the multidisciplinary knowledge of mechanical design, optics, signal processing, pattern recognition and intelligent control to realize fast intelligent analysis and processing of paper money.

In early 2000, the research group launched the theory of banknote analysis and its application technology. In 2003, the first banknote clearing machine was developed in China. In the process of equipment development, the project group has made breakthroughs and many innovative achievements in the key technology of many modern financial optical and electronic equipment, such as the structure design of the paper currency operation channel and its intelligent control technology, the multi spectral sensor design technology, the paper currency anti-counterfeiting information collection technology, the multi mode paper money data comprehensive processing and analysis technology and so on. The equipment produced by this technology is the only product approved by the Ministry of public security. Compared with the similar products at home and abroad, the performance of the paper currency sorting is more stable, the authenticity of the authenticity identification is more reliable, and the continuous operation ability of the equipment is stronger.

The research group and the people's Bank of China and the Ministry of Public Security jointly formulated the national standard for general technical conditions of the RMB discriminator. There are five patents, five software copyrights, 22 academic papers published and 27 cited.

The Harbin film studio has applied this technology to batch production of CF series banknote clearing machine. These devices have been used by financial institutions at all levels in China, and the whole machine and core components have been exported to 20 countries in the world. The long term cooperation between the project group and the Harbin film machine works, giving full play to their respective advantages, making our province a production base of modern financial equipment, and also promoting the development of machinery processing, electronics, software and other related industries in our province. As of December 2008, the technology created a direct output value of one hundred and fifty million.