Prof. NGAN King Ngi, IEEE & IET Fellow
University of Electronic Science and Technology of China, China
Speech Title: Object Segmentation and Its Visual Quality Assessment for Images
King N. Ngan (F’00) received the Ph.D. degree in Electrical Engineering from the Loughborough University in U.K. He is currently a Chair Professor at the University of Electronic Science and Technology, Chengdu, China, under the National Thousand Talents Program. He was previously a Chair Professor at the Chinese University of Hong Kong, the Nanyang Technological University, Singapore, and the University of Western Australia, Australia. He holds honorary and visiting professorships of numerous universities in China, Australia and South East Asia.
Prof. Ngan served as associate editor of IEEE Transactions on Circuits and Systems for Video Technology, Journal on Visual Communications and Image Representation, EURASIP Journal of Signal Processing: Image Communication, and Journal of Applied Signal Processing. He chaired and co-chaired a number of prestigious international conferences on image and video processing including the 2010 IEEE International Conference on Image Processing, and served on the advisory and technical committees of numerous professional organizations. He has published extensively including 3 authored books, 7 edited volumes, over 400 refereed technical papers, and edited 9 special issues in journals. In addition, he holds 15 patents in the areas of image/video coding and communications.
Prof. Ngan is a Fellow of IEEE (U.S.A.), IET (U.K.), and IEAust (Australia), and an IEEE Distinguished Lecturer in 2006-2007.
Abstract: Object segmentation is a key technique to extract objects in an image to aid the analysis and processing of its content. Therefore, it is necessary to develop good object segmentation algorithms for generating accurate segmentation results. Object segmentation is often a pre-processing step for different applications which have different quality requirements. Current commonly used statistics-based object segmentation quality assessment methods are not able to satisfy all applications. Therefore, specific quality assessment methods should be designed for the intended application.
In the first part, we investigate interactive object segmentation with an input rectangle. Here, a coarse-to-fine method from region-level segmentation to pixel-level segmentation is presented. In the region-level segmentation, the best combination of adjacent refined superpixels is selected as the coarse segmentation result by measuring its global contrast and tightness degree. Subsequently, we use the coarse segmentation result to construct the energy function in the pixel-level segmentation. The result can be further refined with the fusion of the region-level and pixel-level segmentation.
In the second part, we explore visual quality assessment of object segmentation in terms of subjective evaluation and objective measure. Firstly, we present a subjective object segmentation visual quality database, in which a total of 255 segmentation results were evaluated by more than 30 human subjects. This database is used to evaluate the performance of the objective measures and analyze their pros and cons. Then, we propose a full-reference objective measure for object segmentation visual quality evaluation, which involves four human visual properties. Finally, our measure is compared with some state-of-the-art objective measures on our database. The experiment demonstrates that the proposed measure performs better in matching subjective judgments.
Prof. David Zhang, IEEE & IAPR Fellow
Hong Kong Polytechnic University, Hong Kong
Speech Title: Advanced Biometrics
David Zhang graduated in Computer Science from Peking University. He received his MSc in 1982 and his PhD in 1985 in both Computer Science from the Harbin Institute of Technology (HIT), respectively. From 1986 to 1988 he was a Postdoctoral Fellow at Tsinghua University and then an Associate Professor at the Academia Sinica, Beijing. In 1994 he received his second PhD in Electrical and Computer Engineering from the University of Waterloo, Ontario, Canada. Currently, he is a Chair Professor at the Hong Kong Polytechnic University where he is the Founding Director of Biometrics Research Centre (UGC/CRC) supported by the Hong Kong SAR Government in 1998. He also serves as Visiting Chair Professor in Tsinghua University and HIT, and Adjunct Professor in Shanghai Jiao Tong University, Peking University, National University of Defense Technology and the University of Waterloo. He is the Founder and Editor-in-Chief, International Journal of Image and Graphics (IJIG); Book Editor, Springer International Series on Biometrics (KISB); Organizer, the first International Conference on Biometrics Authentication (ICBA); Associate Editor of more than ten international journals including IEEE Transactions and so on. So far, he has published over 20 monographs, 400 international journal papers and 40 patents from USA/Japan/HK/China. He has been continuously listed as a Highly Cited Researchers in Engineering by Clarivate Analytics (formerly known as Thomson Reuters) in 2014, 2015, 2016 and 2017, respectively. Professor Zhang is a Croucher Senior Research Fellow, Distinguished Speaker of the IEEE Computer Society, and a Fellow of both IEEE and IAPR.
Abstract: The market for artificial intelligence (AI) technologies is flourishing. As one of the important AI technologies, biometrics has been an area of particular interest. It has let to the extensive study of biometrics technologies and the development of numerous algorithms, applications, and systems, which could be defined as Advanced Biometrics. This talk will be focused on its new biometrics research trend. As case studies, two new biometrics applications (medical and aesthetical biometrics) are explored. Some useful achievements could be given to illustrate their effectiveness.
Prof. Hong Yan, IEEE & IAPR Fellow
City University of Hong Kong, Hong Kong
Speech Title: Online Learning Methods for Human Face Tracking and Facial Expression Recognition
Hong Yan received his PhD degree from Yale University. He was Professor of Imaging Science at the University of Sydney and is currently Chair Professor of Computer Engineering at City University of Hong Kong. His research interests include image processing, pattern recognition and bioinformatics, and he has over 300 journal and conference publications in these areas. Professor Yan was elected an IAPR fellow for contributions to document image analysis and an IEEE fellow for contributions to image recognition techniques and applications. He received the 2016 Norbert Wiener Award from the IEEE SMC Society for contributions to image and biomolecular pattern recognition techniques.
Abstract: Human face image analysis has many applications to security operations, human-machine interactions, e-learning and digital entertainment. In this presentation, a tensor based model recently developed by our research group will be introduced for real-time human face tracking. We formulate a video segment as a third-order tensor and perform incremental singular value decomposition (SVD) to update the object template. The procedure leans a low-rank representation of the tensor. In our system, face tracking is integrated with facial expression recognition. Gabor wavelet features have been proven to be effective for facial expression recognition. However, there are too many features if we consider multiple scales and orientations of the wavelets. We have recently developed a co-clustering based method to solve this problem. In the co-clustering process, we can extract most discriminant features and eliminate irrelevant ones. Experiments on several video and image databases demonstrate that our system can recognize seven types of common facial expressions, neutral, happy, sad, surprise, anger, disgust and fear, with high accuracy.
Prof. Yulin Wang
Wuhan University, China
Speech Title: Image Authentication and Tamper Localization based on Semi-Fragile Hash Value
Prof. Yulin Wang is a full professor and PhD supervisor in International School of Software, Wuhan University, China. He got PhD degree in 2005 in Queen Mary, University of London, UK. Before that, he has worked in high-tech industry for more than ten years. He has involved many key projects, and hold 8 patents. He got his master and bachelor degree in 1990 and 1987 respectively from Xi-Dian University, and Huazhong University of Science and Technology（HUST）, both in China. His research interests include digital rights management, digital watermarking, multimedia and network security, and signal processing. In recently 10 years, Prof. Wang has published as first author 3 books, 40 conference papers and 45 journal papers, including in IEEE Transactions and IEE proceedings and Elsevier Journals. Prof. Wang served as editor-in-chief for International Journal of Advances in Multimedia in 2010. He served as reviewer for many journals, including IEEE Transactions on Image Processing, IEEE Signal Processing Letters, Elsevier Journal of Information Sciences. He served as reviewer for many research funds, including National High Technology Research and Development Program of China ( ‘863’ project). Prof. Wang was the external PhD adviser of Dublin City University, Ireland during 2008-2010. He was the keynote speakers in many international conferences. He bas been listed in Marcus ‘who’s who in the world’ since 2008.
Abstract: Image authentication can be used in many fields, including e-government, e-commerce, national security, news pictures, court evidence, medical image, engineering design, and so on. Since some content-preserving manipulations, such as JPEG compression, contrast enhancement, and brightness adjustment, are often acceptable—or even desired—in practical application, an authentication method needs to be able to distinguish them from malicious tampering, such as removal, addition, and modification of objects. Therefore, the traditional hash-based authentication is not suitable for the application. As for the semi-fragile watermarking technique, it meets the requirements of the above application at the expense of severely damaging image fidelity. In this talk, we propose a hybrid authentication technique based on what we call fragile hash value. The technique can blindly detect and localize malicious tampering, while maintaining reasonable tolerance to conventional content-preserving manipulations. The hash value is derived from the relative difference between each pair of the selected DCT AC coefficient in a central block and its counterpart which is estimated by the DC values of the center block and its adjacent blocks. In order to maintain the relative difference relationship when the image undergoes legitimate processing, we make a pre-compensation for the AC coefficients. Experimental results show that our technique is superior to semi-fragile techniques, especially in image fidelity, tolerance range of legitimate processing, and/or the ability to detect and localize the tampered area. Due to its low computational cost, our algorithm can be used in real-time image or video frame authentication. In addition, this kind of proposed techniques can be extended to use other characteristic data, such as high-level moment, statistical data of image, and so on.
Assoc. Prof. Ray C.C. Cheung
City University of Hong Kong, Hong Kong
Speech Title: Algorithm and Architecture for Intra Prediction of HEVC
Dr. Ray Chak-Chung Cheung received the B.Eng.(Hons) and M.Phil. degrees in computer engineering and computer science & engineering from The Chinese University of Hong Kong (CUHK) in 1999 and 2001 respectively, and the DIC and Ph.D. degree in computing from Imperial College London (IC) in 2007.
He joined the Department of Computer Science & Engineering at the Chinese University of Hong Kong (CUHK) in 2002 as an Instructor. Before that, he worked as a system administrator in a parallel cluster computing company, Cluster Technology Limited for one year. Two years later in 2004, he received theHong Kong Croucher Foundation Scholarship and moved to London where he spent three years for his Ph.D. study. He was with Stanford University and UCLA in the summer of 2005 and 2006 as a visiting scholar. After completing his Ph.D. study, he received the Hong Kong Croucher Foundation Fellowship and moved to Los Angeles, in the Electrical Engineering department at UCLA, where he spent two years with Image Communication Lab for continuing his research work. In 2008, he took a 6-month internship at a top IC design company in Hong Kong, Solomon-Systech Limited as a senior digital IC designer. In 2009, he visited the PALMS group in the Department of Electrical Engineering at Princeton University as a visiting research fellow before returning to Hong Kong. He is currently an Associate Professor in the Department of Electronic Engineering at City University of Hong Kong, and with Digital Systems Lab. His current research interests include cryptographic hardware designs and design exploration of System-on-Chip (SoC) designs and embedded system designs. He is the chairman of the IEEE Hong Kong Section CAS/COM Chapter, and executive committee member of the IEEE Hong Kong Section Computer Chapter. He is the author of over 100 journal papers and conference papers.
Prof. Lih-Shyang Chen
National Cheng Kung University, Taiwan
Speech Title: Display and Manipulation of 2D and 3D Medical Images for Anatomy Education
Prof. Lih-Shyang Chen is a Professor of Department of Electrical Engineering of National Cheng Kung University, Taiwan. He received his PhD from University of Pennsylvania in 1987. He had worked as a researcher at Bell Laboratories, USA in 1987-1990 and Director of Computer Center, Ministry of Education, Taiwan in 1996-2000. His research interests include Computer Graphics and Virtual Reality, Image Processing and Computer Vision, Internet Network Systems and Internet Web applications, Multimedia Application System.
Abstract: Human anatomy is the basic scientific study for medical students to learn the shape, position, size, and various relationships of the organ structures in the human body. The Visible Human Project (VHP)– a high resolution digital collection of medical image data of a single human specimen is a large digital archive of medical image data of the US National Library of Medicine. In this presentation, we make use of the images produced by the VHP to develop a 2D and 3D anatomy learning system for the students in medical schools to learn human anatomy more effectively and efficiently. To this end, we developed an interactive segmentation system that allows the users to segment objects of interest from the original images and build the 3D organ models. Once the 3D organ models are constructed, the users can interactively display 3D anatomy objects and related 2D images to visualize the details of 3D anatomy objects and their spatial relations in a virtual reality environment. We also describe some functions that can facilitate the medical education through some e-learning tools.
Assoc. Prof. Kin Hong Wong
The Chinese University of Hong Kong, Hong Kong
Speech Title: Image and sensor based pose tracking and applications
Professor Wong Kin Hong is an Associate Professor of the Department of Computer Science and Engineering of the Chinese University of Hong Kong. He received a PhD from the Department of Engineering of the University of Cambridge. His major research interest is in 3-D computer vision especially in pose estimation, structure from motion and tracking. He has investigated and developed many useful techniques in computer vision such as the four-point pose estimation algorithm and Kalman-trifocal pose estimation methods which are useful in many application areas such as automatic driving and virtual reality. He is also interested in pattern recognition, embedded applications, and computer music.
Abstract: In this talk, I will concentrate on pose estimation and structure from motion approaches to be applied to multimedia applications. Traditional approaches, such as linear and non-linear methods will be mentioned, and different pose tracking schemes are explained. The use of a machine learning method of applying neural networks for pose estimation is also discussed in this seminar. Furthermore, I will introduce the idea of how to build a sensor fusion system of using an Inertial Measurement Unit (IMU) and vision together. The advantage of this approach is that it can achieve high sampling rate and drift free pose tracking. Finally, I will talk about a multimedia computer vision project of designing a language assistant for visitors to understand foreign texts when traveling abroad.