LeiLei Gao

Leilei Gao (멕잉잉)

Hello!
I am a PhD studnet at the School of Electronics Engineering and Computer Science at Peking University. I joined the Computer Graphics and Interactive Technology Lab in 2008, supervised by Guoping Wang. My research has mainly focused on digital geometry processing, a fascinating subfield of graphics. My current interests include shape analysis and construct models from space curves. For further details about my background, please take a look at my CV.

I was borned in Baoding, a city of Hebei province, People's Republic of China, bordering the national capital Beijing. In my free time, I enjoy jogging and hiking in the mountains. I have finished my first half marathon in Beijing Marathon 2011, within 2 hours and 15 minutes. Now, I'm wishing for my full marathon.

Research

	Create Blending Shapes from Section Curves We present a method that creates blending shapaes from several section curves. Existing methods seldom consider the blending of sharp features. To solve this problem, we detect prominent feature points in section curves, and use an optimization method to find correspondences between those features. Then, we use several meaningful parameters to control the potential blending curves between features. With an initial blending is created by an existing method, we use the blending between features to guide the optimization of the initial blending. According our curent experiments, this method seems to be able to create desirable blending shapes with controllable blendings of sharp features. Also some optimization is used to get the final parts.
	Create Surfaces from Space Curves CurveNetSurf: Creating Surfaces from Curve Networks. Leilei Gao, Lifeng Zhu and Guoping Wang. The 12th International CAD/Graphics 2011 conference Creating Surfaces from Curve Networks. Leilei Gao, Lifeng Zhu and Guoping Wang. Journal of Computer Science and Technology(recommeded, under review)
	Mesh Segmentation Using Learning Methods Segmentation is fundamental to shape understanding and processing. Although there are many existing methods to solve this problem, but few works aim to detect meaningful parts, which is helpful for geometric modeling, manufacturing, animation and texturing of 3d meshes. Prior knowledge is important for meaningful parts detection, considering some intelligent understandings are involved. With labeled training set, we extract several distinguishing features based on diffusion kearnels, then train a classifier, to detect the meaningful parts.

Photo Collections

An image is suppoed to show up here...

MiaoFeng Mountain

An image is suppoed to show up here...

Natural Park of Phoenix Hill