Todd Zickler
Assistant Professor of Electrical Engineering
Division of Engineering and Applied Science
Harvard University

email: zickler {at} eecs.harvard.edu


home research publications courses contact CV


New in 2008:
Autotagging Facebook: Social Network Context Improves Photo Annotation to be presented at the Internet Vision Workshop 2008. (with Zak Stone and Trevor Darrell). [Preprint]
A Perception-based Color Space for Illumination-invariant Image Processing to be presented at SIGGRAPH 2008. (With Hamilton Chong and Steven Gortler). [Preprint]
What do Color Changes Reveal about an Outdoor Scene? to be presented at CVPR 2008. (With Kalyan Sunkavalli, Fabiano Romeiro, Wojciech Matusik and Hanspeter Pfister). [Preprint].
Color Constancy Beyond Bags of Pixels to appear at CVPR 2008. (With Ayan Chakrabarti and Keigo Hirakawa). [Preprint]
Dense Specular Shape from Multiple Specular Flows to appear at CVPR 2008. (With Yuriy Vasilyev, Yair Adato, and Ohad Ben-Shahar). [Preprint].
Photometric Stereo with Non-parametric and Spatially-varying Reflectance to appear at CVPR 2008. (With Neil Alldrin and David Kriegman). [Preprint].
Color Subspaces as Photometric Invariants accepted for publication in IJCV. (With Satya Mallick, David Kriegman and Peter Belhumeur). [Preprint].

2007:
Toward a Theory of Shape from Specular Flow. ICCV 2007. (With Yair Adato, Yuriy Vasilyev, and Ohad Ben-Shahar). [PDF].
The von Kries Hypothesis and a Basis for Color Constancy. ICCV 2007. (With Hamilton Chong and Steven Gortler). [PDF].
Isotropy, Reciprocity and the GBR Ambiguity. CVPR 2007. (With Ping Tan, Satya Mallick, David Kriegman, and Long Quan). [PDF].


Some of this work is part of an ongoing project on appearance capture that is being funded by an NSF CAREER award. For more details, see the project page.


SUV Space

Photometric Invariants from Color Subspaces: Pure (specular-free) diffuse information by simple projections in color space. Applications include stereo, optical flow, shape-from-shading, segmentation, etc.

<--MOUSE OVER

Specular/Diffuse Separation

Specular/Diffuse Separation: Separating reflection components using multi-scale erosion. Applications to dichromatic editing.


<--MOUSE OVER

Reflectance Sharing

Reflectance Sharing: Image-based rendering from a sparse set of images.

Helmholtz Stereopsis

Helmholtz Stereopsis: Recovering the shape of surfaces with non-specular, specular or arbitrary surface reflectance.

Updated: June 28, 2006