My main research thrust is high-level imaging science. From biomedicine to recreational video, imaging data is ubiquitous. Yet, imaging scientists and intelligence analysts are without an adequate language and set of tools to fully tap the information-rich image and video. I work to provide such a language; specifically, I study the coupled problems of segmentation and recognition from a Bayesian perspective emphasizing the role of statistical models in efficient visual inference. My long-term goal is a comprehensive and robust methodology of automatically mining, quantifying, and generalizing information in large sets of projective and volumetric images and video. The following four questions drive my current research inquiries:

1. How to use principled hierarchical structures to model complex real-world phenomena?
2. How to handle the massive data glut for machine learning yet require little or no user labeling?
3. How to incorporate prior high-level knowledge (semantics, context, etc.) during both learning and inference?
4. How to understand and enhance the role of the user in active semi-supervised learning scenarios?

More broadly, my research interests are in the fields of computer vision and medical imaging, machine intelligence, statistical learning, perceptual interfaces and smart environments.

NSF CAREER project is underway. From representation to learning to inference, effective use of high-level semantic knowledge in computer vision remains a challenge in bridging the signal-symbol gap. This research investigates the role of semantics in visual inference through the generalized image understanding problem: to automatically detect, localize, segment, and recognize the core high-level elements and how they interact in an image, and provide a parsimonious semantic description of the image. Specifically, this research examines a unified methodology that integrates low- (e.g., pixels and features), mid- (e.g. latent structure), and high-level (e.g., semantics) elements for visual inference. Adaptive graph hierarchies induced directly from the images provide the core mathematical representation. A statistical interpretation of affinities between neighboring pixels and regions in the image drives this induction. Latent elements and structure are captured with multilevel Markov networks. A probabilistic ontology represents the core knowledge and uncertainty of the inferred structure and guides the ultimate semantic interpretation of the image. At each level, rigorous methods from computer science and statistics are connected to and combined with formal semantic methods from philosophy.

This part of the site contains information regarding my (sometimes too) broad research program. To cater to different audiences who may be using the site, I have organized the various pages into four groups, which are linked on the left bar. The first group is based on active project, which is usually tied to a grant or contract. The second group is based on technical problem/contribution, which is probably most useful for peer researchers in the field. The third group is based on specific application; many of the ideas from the technical problem section will have been used in sovling the various applications. The fourth group is based on funding agency, but the pages linked there simply list the various grants (with links) to the actual project pages. Note some information on the various pages is redundant; this is intentional.

To return to this main page, you can click on the "Research Pages" link on the top of the left bar.

I've spent a good deal of time deliberating how to best organize these pages; if you have comments, I would be happy to hear them.

We gratefully acknowledge funding support from the following agencies and programs.