Ganesh Ramanarayanan's Webpage

My research interests are fairly broad, spanning graphics, visual perception, parallel computing, and artifical intelligence. During the last few years, my work has focused on:

• Perceptually based and feature-based rendering algorithms
• Texture synthesis, detail synthesis, and image super-resolution
• Generalized parallelization frameworks for multicore machines

My resume is available here.

Projects and Publications

			Perception of Complex Aggregates SIGGRAPH 2008 (conditionally accepted) An investigation of how aggregate geometry is perceived, in particular, how aggregate appearance is affected when one kind of object is replaced with another.
			Scheduling Strategies for Optimistic Parallel Execution of Irregular Programs SPAA 2008 A framework for flexible scheduling policies when parallelizing irregular, data-dependent programs.
			Dimensionality of Visual Complexity in Computer Graphics Scenes SPIE HVEI 2008 Through a psychophysical study and multidimensional scaling analysis, we analyze visual complexity in graphics for a representative collection of renderings.
			Optimistic Parallelism Benefits from Data Partitioning ASPLOS 2008 Partitioning is often used to efficiently parallelize applications on structured data. Here we show how to partition effectively for optimistic parallelization of irregular applications.
			Visual Equivalence: Towards a New Standard for Image Fidelity SIGGRAPH 2007 Visual equivalence occurs when noticeably different images convey the same scene appearance to an observer. This paper analyzes this phenomenon for illumination transformations and proposes a metric for novel rendering optimizations.
			Optimistic Parallelization Requires Abstractions PLDI 2007 Parallelize irregular applications, such as mesh refinement and clustering, by defining commutativity conditions on shared data structure methods.
			Constrained Texture Synthesis via Energy Minimization IEEE TVCG 2007 Jan/Feb, Vol. 13, No. 1 Control texture synthesis using constraint images and graphcut energy minimization, resulting in a fast, robust, and high-quality algorithm that improves the viability of analogies-based synthesis frameworks.
			Feature-Based Textures EGSR 2004 Add explicit discontinuity features to images in order to improve reconstruction during arbitrary rescaling in texture mapping.
			Publication/Citation: A Proof-Theoretic Approach to Mathematical Knowledge Management Cornell University TR2005-1985 A formal reasoning system capturing the mechanism of theorem reuse in mathematics.

Links

• Recent research in image analysis and synthesis

Personal

• Music is one of my biggest hobbies. I am a student of Carnatic (South Indian classical) music, training primarily in vocal and percussion (on an instrument called the mridangam). I have performed frequently in the NJ area. After coming to Cornell, I have organized several concerts under the auspices of SPICMACAY. Recently I started an East-West fusion band with some other Cornell students called Moksha (link broken). Our first album is available for free download. I highly recommend our original instrumental tracks, Revival and Fusion 2 (so named because it was our second fusion song). I also have some training in Western classical music theory and flute.
• Foreign language related stuff.
• Pictures that have (not so) recently resided on my camera. I haven't been able to update this for a long time. Presumably this is because I have been working hard (at least that's my version of the story)
• In sixth grade I was on the PBS kids game show "Where in the World is Carmen Sandiego". I did not win the free trip, but I did win a sleuth jacket (picture soon) and a camera. Running across a map of South America is harder than it looks.