Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
SIGGRAPH 2007: Microsoft Research Leads the Way
By Rob Knies
August 6, 2007 12:01 AM PT

These days, with the mushrooming popularity of digital cameras and camera phones, everybody’s a photographer. But that doesn’t mean that everybody’s a good photographer.

One of the most vexing problems confronted by amateur shutterbugs is the desire to take a photograph under dim lighting conditions. Such scenarios offer a couple of strategies: 1) shoot the photo using a long exposure time, or 2) use a high ISO setting to increase the light sensitivity of the camera’s sensor.

There are, of course, often-undesirable tradeoffs with each approach. The former, used with a handheld camera, can induce a blurred image from camera movement. The latter adds unwanted artifacts to an image due to “noise” from a lack of illumination.

What’s a digital photographer, more enthusiast than expert, to do? A solution may be revealed during SIGGRAPH 2007.

The event, the 34th International Conference and Exhibition on Computer Graphics and Interactive Technologies, will be held Aug. 5-9 at the San Diego Convention Center, and one of the select highlights from the SIGGRAPH 2007 papers program is called Image Deblurring with Blurred/Noisy Image Pairs.

That paper, co-authored by Lu Yuan of The Hong Kong University of Science and Technology, Jian Sun of Microsoft Research Asia, Long Quan of The Hong Kong University of Science and Technology, and Harry Shum of Microsoft Research Asia, explains a technique that combines a blurred image from the first scenario with a noisy image from the second to produce a high-quality photograph.

Such advances are one of the reasons why Microsoft Research continues to have such a significant presence in the annual SIGGRAPH conferences. Of the 108 papers accepted for this year’s gathering, 14 (13 percent) are from Microsoft Research, more than any other organization.

Such contributions represent the extension of a trend. Microsoft Research supplied 10 percent of the accepted SIGGRAPH papers in 2002, 13 percent in 2003, 12 percent in 2004, 19 percent in 2005, and almost 20 percent in 2006.

Furthermore, Microsoft Research’s SIGGRAPH participation underscores an ongoing emphasis on collaboration with academia. Of the 14 papers the organization had accepted for this year’s conference, 11 represent collaborations with university partners.

Researchers from three of Microsoft Research’s five worldwide labs—those in Asia, the United Kingdom, and Redmond—had papers accepted for SIGGRAPH 2007.

Microsoft Research papers accepted by SIGGRAPH 2007 (authors from Microsoft Research unless otherwise noted):

Capturing and Viewing Gigapixel Images
Johannes Kopf, University of Konstanz; Matt Uyttendaele; Oliver Deussen, University of Konstanz; Michael F. Cohen.

Design of Tangent Vector Fields
Matthew Fisher, California Institute of Technology; Peter Schröder, California Institute of Technology; Mathieu Desbrun, California Institute of Technology; Hugues Hoppe.

Direct Manipulation of Subdivision Surfaces on GPUs
Kun Zhou, Xin Huang, Weiwei Xu, Baining Guo, Heung-Yeung Shum.

Efficient Symbolic Differentiation for Graphics Applications
Brian K. Guenter.

Image-Based Tree Modeling
Ping Tan, The Hong Kong University of Science and Technology; Gang Zeng, The Hong Kong University of Science and Technology; Jingdong Wang, The Hong Kong University of Science and Technology; Sing Bing Kang; Long Quan, The Hong Kong University of Science and Technology.

Image Deblurring with Blurred/Noisy Image Pairs
Lu Yuan, The Hong Kong University of Science and Technology; Jian Sun; Long Quan, The Hong Kong University of Science and Technology; Heung-Yeung Shum.

Image Vectorization Using Optimized Gradient Meshes
Jian Sun; Lin Liang; Fang Wen; Heung-Yeung Shum.

Interactive Relighting With Dynamic BRDFs
Xin Sun, Zhejiang University; Kun Zhou; Yanyun Chen; Steve Lin; Jiaoying Shi, Zhejiang University; Baining Guo.

Joint Bilateral Upsampling
Johannes Kopf, University of Konstanz; Michael F. Cohen; Dani Lischinski, The Hebrew University; Matt Uyttendaele.

Gradient Domain Editing of Deforming Mesh Sequences
Weiwei Xu; Kun Zhou; Yizhou Yu, University of Illinois at Urbana-Champaign; Qifeng Tan, Zhejiang University; Qunsheng Peng, Zhejiang University; Baining Guo.

Mesh Puppetry: Cascading Optimization of Mesh Deformation With Inverse Kinematics
Xiaohan Shi, Zhejiang University; Kun Zhou; Yiying Tong, California Institute of Technology; Mathieu Desbrun, California Institute of Technology; Hujun Bao, Zhejiang University; Baining Guo.

Photo Clip Art
Jean-François Lalonde, Carnegie Mellon University; Derek Hoiem, Carnegie Mellon University; Alexei A. Efros, Carnegie Mellon University; Carsten Rother; John Winn; Antonio Criminisi.

ShapePallettes: Interactive Normal Transfer Via Sketching
Tai-Pang Wu, The Hong Kong University of Science and Technology; Chi-Keung Tang, The Hong Kong University of Science and Technology; Michael S. Brown, Nanyang Technological University; Heung-Yeung Shum.

Soft Scissors: An Interactive Tool for Real-Time, High-Quality Matting
Jue Wang, University of Washington; Maneesh Agrawala, University of California, Berkeley; Michael F. Cohen.