INVITED SPEAKERS 特邀嘉宾
Prof. Nobuhiko Mukai
Tokyo City University, Japan
Nobuhiko Mukai is a professor
of Graduate School of Integrative Science and Engineering,
Tokyo City University. He received his B.E., M.E., and Ph.D
degrees from Osaka University in 1983, 1985, and 2001
respectively. He started to work at Mitsubishi Electric
Corporation and changed to work as an associate professor at
Musashi Institute of Technology in 2002. He is currently a
professor of Tokyo City University from 2007. His research
interests are computer graphics and image processing. He is
a member of ACM, SAS, VRSJ, IEICE, ITE, IPSJ, IIEEJ, and
Speech Ttile: Computer Graphics Applications with Particle Methods
Abstract: In order to visualize natural phenomena such as aurora, rainbow, lightning, and avalanche, physics based simulation is necessary. There are some methods to simulate the behavior of continuous body. One of them is particle method, which is robust for topological change such as separation and destruction. In this speech, three applications using particle methods are introduced: viscoelastic fluid, waterfall, and blood. Viscoelastic body has two characteristics of viscosity and elasticity, and there is no formulated constitutive equation. Then, we have used a new constitutive equation to simulate the behavior. The next talk is about waterfall, which has large scale from the lip to the basin. We have divided the model into three parts: water stream, splashing spray, and spray cloud, and applied three different equations to simulate each behavior. Finally, blood flow simulation in the heart is presented. In the heart, there is a valve called aortic valve. If it falls into malfunction, surgeries are operated, and computer simulation is necessary before the surgery. In this talk, blood flow simulation from the left ventricle to the aorta is visualized with particles and the pressure change in the simulation is compared to the real data.
Assoc. Prof. Hongjun Li
Beijing Forestry University, China
Hongjun Li received the MS degree in Mathematics from the University of Science and Technology, Beijing (USTB), in 2003 and the PhD degree in Computer Application from the Institute of Automation, Chinese Academy of Sciences (CASIA), in 2012. He is currently Dean of the Department of Mathematics and an associate professor in Beijing Forestry University (BJFU). His research interests include geometry modeling, computer graphics, virtual reality and image processing. He is (co-)author of over 40 papers published in journals and conference proceedings. He holds a U.S. patent and eight Chinese patents. He is a member of China Computer Federation and a member of the Council of Beijing Mathematics Association.
Speech Ttile: Analysis and Reconstruction of 3D Point Cloud
Abstract: Three-dimensional (3D) point cloud data can be obtained by laser scanning. Such data provide real 3D information that reflects the real shape of an object with high precision and high resolution. Analysis and reconstruction of 3D point cloud are helpful to the digitalization of a city, virtual scenes, 3D printer and automatic driven. In this speech, we will introduce three aspects of the work: (1) A method called Chord And Normal vectors (CAN) has been proposed for estimating the principal curvatures and the principal directions; (2) Both 3D point cloud classification and segmentation are illustrated with our new algorithms; (3) We present a tree reconstruction approach based on scan point cloud.
Prof. Qiu Chen
Kogakuin University, Japan
Qiu Chen received Ph.D. degree in electronic engineering from Tohoku University, Japan, in 2004. Since then, he has been an assistant professor and an associate professor at Tohoku University. He is currently a professor at Kogakuin University. His research interests include pattern recognition, computer vision, information retrieval and their applications. Prof. Chen serves on the editorial boards of several journals, as well as committees for a number of international conferences.
Speech Ttile: Recent Advancements in Scene Recognition Research
Abstract: In recent years, deep learning approaches have won almost all contests in pattern recognition and machine learning areas, especially the object recognition task can be approximately addressed because the accuracy has achieved non-trivial performance compared with results in previous works. However, the performance of scene recognition is still far away from the desirable extent due to more complex constitutions.
Over the past several years, the methods of scene recognition have undergone a vital evolution as with the developments of computer vision and deep learning. The purpose of this paper is to review current popular and effective means of scene recognition, which is expected to bring benefits to future research or practical applications. We seek to reveal the relationships among different approaches and discover the critical components that leading to remarkable performance.
Through the analysis of some representative schemes, the motivation and insights behind them will be uncovered, which will help to facilitate the design of better scene recognition approaches. Furthermore, potential problems and promising directions are discussed.