MIT thesis defense: Learning Continuous Models for Estimating Intrinsic Component Images

Speaker: Marshall Tappen , MIT CSAIL
Date: Tuesday, May 16 2006
Time: 10:30AM to 11:30AM

Interpreting an image of a scene is difficult because the various characteristics of the scene contribute to its appearance. For example, an edge in an image could be caused by either an edge on a surface or change in the surface's color. Distinguishing the effects of different scene characteristics is an important step towards high-level analysis of an image.

This talk will describe how to use machine learning to build a system that recovers different characteristics of the scene from a single, gray-scale image of the scene. Using the observed image, the system estimates a shading image, which captures the interaction of the illumination and shape of the scene pictured, and an albedo image, which represents how the surfaces in the image reflect light. Measured both qualitatively and quantitatively, this system produces state-of-the-art estimates of shading and albedo images. This system is also flexible enough to be used for the separate problem of removing noise from an image.

Building this system requires algorithms for continuous regression and learning the parameters of a Conditionally Gaussian Markov Random Field. Unlike previous work, this system is trained using real-world surfaces with ground-truth shading and albedo images.

Committee Members:
Professor Edward Adelson
Professor William Freeman
Professor Michael Collins