The Center for Research on Intelligent Perception and Computing
     Chinese
Research  
   Research focus
Research focus

Biometrics
tnt group
 
We have started biometrics research since 1998 and contributed greatly to iris, face (both 2D and 3D), fingerprint, palmprint, palm vein, handwriting, gait recognition and multi-biometrics areas. The group undertakes projects from to National Basic Research Program of China, National Sci. & Tech. Support Program, Instrument Developing Project of the Chinese Academy of Sciences, National Natural Science Foundation of China and other research projects. It aims to combine the imaging techniques and identification algorithms to enhance the users’ capacity adaptable to the environment in the multi-modal biometric recognition system, and to accomplish iris and face recognition from the controlled scenes to complex scenes. The group will insist on the innovation of obtaining the features of face and iris under complex scenes. Furthermore, the group will follow the technology roadmap: “imaging devices - data resources - recognition algorithms - integrated systems - practical applications”, and provide high-level identification technology for public safety. Till now, the group published more than 100 international journal and conference articles and has more than 20 granted patents. Besides, the group was awarded the 2nd Prize of National Technological Invention. Furthermore, the biometric group has released the largest iris dataset in the world and the dataset is used by more than 10,000 research units from in more than 120 countries.
 
 
Image and Video Analysis
tnt group  
Image and video analysis refers to automatically analyzing objects and high-level semantic information from large-scale images and videos based on the techniques of pattern recognition, artificial intelligence and computer vision. The major research directions include: (1) Research on object detection and recognition in images and videos. Against the background of big data, we research the problem of hierarchical and structural visual representation in large-scale object detection and recognition to improve their performance in real applications. (2) Research on object tracking and behavior analysis in wide-area and complex situations. To serve the needs of public security, semantic information of human actions and behaviors in visual surveillance is studied to realize the abnormal behavior detection and analysis. (3) Research on pedestrian attribute analysis in complex camera networks. This group focuses on pedestrian attribute analysis and re-identification in non-overlapping camera networks by exploiting the relationships across cameras.
The group has published many papers in top international conferences and journals, with the second prize of National Scientific and Technological Progress in 2011. In the PASCAL VOC challenge, the group won the first prize of object detection in 2010 and 2011. The group also attended the ImageNet Classification and Localization with additional training data Challenge and won the first prize of image classification in 2014.
 
Big Data and Multi-modal Computing
tnt group  
The big data and multi-modal computing research group deals with different types of data, such as images, text, videos, etc. It studies both theory and applications about pattern recognition, visual computing, machine learning, data mining, context modeling, etc. The main achievements include the champion of 2013 Internet Contest for Cloud & Mobile Computing in human segmentation, the best student paper award of ICPR2014, etc. Major research interests of the multi-modality computing group include: (1) Visual computing based on non-Euclidean space and large scale and multi-modality data analysis. It studies the computational theories and mechanisms on topology-based object representations, unsupervised clustering, cross-modality data analysis and large scale machine learning. (2) Methods and applications for visual computing based on deep learning. It studies how to efficiently integrate the feedback mechanism in feedforward networks and how to combine active vision in feedforward and feedback networks, which can be used to solve many vision tasks in large scale vision analysis, e.g., object recognition, object detection, video segmentation and video analysis. (3) Intelligent data analysis for public security and business intelligence. It studies the key technology of large scale social network data mining taking advantage of advanced technologies of big data, such as context modeling, time-series prediction and user modeling, which adapts to the needs of public and content securities.
 
Content Security and Authentication
tnt group  
The group focuses on cyber content security and authentication, by analyzing the integrality and authenticity for the image content, based on pattern recognition and statistical analysis tools. The research of cyber content security is very important for information forensics and security. The research topic includes: (1) Research on the statistic feature model of image forensics. (2) Feature learning based image forensics. (3) Image tempering detection and localization based on pattern analysis. (4) Source classification and device linking and (5) Research on prototype systems of image forensics.
 
Sensing and Information Acquisition
tnt group  
The sensing and information acquisition group will focus on the innovation of intelligent sensing theory, methods and equipment by the cross-fusion of artificial intelligence, cognitive science, optics, electronics and control engineering, and provide the novel technology for National public security and information industry development. The group aims to perceive and describe multi-modal scenes information precisely and achieve accurate reconstruction of the complex scenes. Till now, the group focuses on the research on multi-modal biometric acquisition device, which can capture iris, face and gait biometrics at a distance in the unconstrained environment. Such research achievement will be useful for biometrics research and public security applications. Key technologies include light field computational imaging, intelligent human-computer interaction, multi-biometric data preprocessing, coding, fusion, visualization and so on.

 
Copyright © Editorial Board of The Center for Research on Intelligent Perception and Computing