Navigating the world around us used to entail paper charts and maps, and the skill to interpret the system of signs and symbols that stood in for the real objects and features of the landscape. Not any more with the arrival of 3D mapping that visualizes the real-world route to your destination.
Behind the opening salvos of simplifying travel for you and me lies an array of complex and lengthy research carried out over the past 30 years. It has taken this time and some of the world's leading minds in computer vision to make such a development possible. Beyond, a host of significant uses now beckon.
Among the global pioneers is Prof Long Quan, Computer Science and Engineering, who has devoted his entire academic career to the advancement of computer vision and 3D reconstruction from images through his own innovative contributions to these scientific areas and the mentoring of next-generation research stars.
"This is the future," said Prof Quan, Founding Director of the HKUST Center for Visual Computing and Image Science. "Your mobile phone will show you 3D images and it will be very convenient for moving around. Think of all the other uses for such maps too. Search and rescue teams would know the exact terrain they faced before they set out. The military could use 3D maps to guide missiles.
"When I started, this was a relatively new field. But we already saw the potential for applications."
For Prof Quan, once a child artist who swapped paint brushes for computer vision, visual images are a fundamental element of life that speak to us directly. He has lived in this world of 3D reconstruction from the time of his graduate studies in France in the mid-1980s, which began with his Master's at Université Henri Poincaré and was followed by a doctorate at the Institut National Polytechnique de Lorraine (INPL) (now both part of the University of Lorraine). The two institutions were leading establishments with laboratories affiliated with CNRS (the French National Center for Scientific Research) and INRIA (the French Institute for Research in Computer Science and Automation). Prof Quan was a member of one of the earliest cohorts of Mainland students to receive a scholarship from the Ministry of Education to study overseas, following his graduation from Northern Jiaotong University (now Beijing Jiaotong University) in 1984.
After earning his PhD in 1989, Prof Quan launched his academic career at INRIA Grenoble, France's key public research institute focused on computational sciences, and contributed significantly to the development of modern 3D computer vision in the 1990s through his outstanding series of 3D reconstruction algorithms including the six-point algorithm and projective reconstruction from multiple views. He joined HKUST in 2001, where he remains fascinated by the potential of the area and expects to stay so in the decades ahead.
Recent advances include reconstruction algorithms and systems, developed together with his research team, to address the challenge of missing detail that arises with present mapping technology. Current techniques use unmanned aerial vehicles and ground-level cars to collect image data, with areas that escape attention needing to be filled in manually. Through Prof Quan's breakthrough, manual inputting can be replaced by automated reconstruction of buildings and landmarks, based on the images collected. This can take the technology to fresh heights of accuracy.
Unsurprisingly, Prof Quan's work has attracted attention from other academics and a wide range of companies, including Google, Nokia, Microsoft, Airbus, Thales, and Mainland China industries. He has over 6,700 citations and an H-index (a recognized measure of productivity and citation impact) of 43, according to Google Scholar. In addition, he has served as an editor and chair in all major computer vision journals and conferences respectively, and in 2011 was a General Chair of the IEEE International Conference on Computer Vision, the premier biennial event in this field globally. “I have been very fortunate to be able to contribute to computer vision and serve this young and dynamic community," he said.
Prof Quan believes his on-going enthusiasm is driven by intellectual curiosity and a desire to see the knowledge he uncovers put to use. Along with applications, such uses encompass teaching and sharing his expertise and passion for discovery with students. He is a devoted educator and has twice been elected a HKUST Best Ten Lecturer in an independent student-organized poll. He is also a dedicated supervisor of graduate students, including Peter Sturm, the first awardee of the Best French PhD Dissertation of the Year in Computer Science (Le Prix de Thèse SPECIF 1998, now known as Le Prix de Thèse Gilles Kahn) and Maxime Lhuillier, who received the ICPR Piero Zamperoni Best Student Paper Award in 2000. His graduate students now hold faculty positions in world-renowned institutions and universities, such as CNRS, INRIA, Peking University, National University of Singapore, and Princeton University.
Prof Quan is now driving advances through the further development of algorithms and systems for 3D mapping. Such work has attracted a constant stream of six-figure funding under various Hong Kong Research Grants Council schemes and from industry grants. Several patents have been awarded and others are pending. In addition, a related HKUST start-up has been launched. The search is continuous for ways to advance through greater accuracy, higher resolution, and faster speed, he noted.
One key project is to map the whole of Hong Kong in 3D. "This is our dream," Prof Quan said. "We have started to undertake the work. However, data capture is expensive. I am now applying for funding through different channels to enable us to reach our goal." Such a survey would assist various industries, including logistics and navigation companies. Urban planners could also benefit.
"Computer vision is different from other computer science areas as you see concrete objects, not only abstract numbers," Prof Quan said. "To me, the fascination of the 3D shape is endless. Its construction, which involves both visual artistic form and geometry, is of constant interest. We still have a great deal to explore in terms of applications but we are moving forward."
A Visionary Career
|Prof Long Quan's pathway to pioneering insights:
- BS in Telecommunication, Northern Jiaotong University, Beijing, 1984
- MS (D.E.A.) in Computer Science, Université Henri Poincaré-CNRS-INRIA, France, 1986
- PhD in Computer Science, INPL-CNRS-INRIA, France, 1989
- CNRS Researcher at INRIA, France, 1990-2001
- Associate Professor, Computer Science, HKUST, 2001-05
- Professor, Computer Science and Engineering, HKUST, 2005-now