Error Correction for High-Dimensional Data via Convex Programming

Finally, we show how these theoretical developments lead to simple, scalable, and robust algorithms for face recognition in the presence of varying illumination and occlusion. The idea is extremely simple: seek the sparsest representation of the test image as a linear combination of training images plus a sparse error term due to occlusion. In addition to achieving excellent performance on public databases, this approach sheds light on several important issues in face recognition, such as the choice of features and robustness to corruption and occlusion.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Electrical and Computer Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Wright, John N.

Contributors dc:contributor

• Ma, Yi

• Engineering, Electronics and Electrical

Language dc:language

eng

Identifier

(MiAaPQ)AAI3395542

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/81151