Optimal linear representations of images for object recognition

Author

Liu, Xiuwen ; Srivastava, Anuj ; Gallivan, Kyle

Author_Institution

Dept. of Comput. Sci., Florida State Univ., Tallahassee, FL, USA

Volume

26

Issue

5

fYear

2004

fDate

5/1/2004 12:00:00 AM

Firstpage

662

Lastpage

666

Abstract

Although linear representations are frequently used in image analysis, their performances are seldom optimal in specific applications. This paper proposes a stochastic gradient algorithm for finding optimal linear representations of images for use in appearance-based object recognition. Using the nearest neighbor classifier, a recognition performance function is specified and linear representations that maximize this performance are sought. For solving this optimization problem on a Grassmann manifold, a stochastic gradient algorithm utilizing intrinsic flows is introduced. Several experimental results are presented to demonstrate this algorithm.

Keywords

Markov processes; Monte Carlo methods; gradient methods; image representation; object recognition; optimisation; Grassmann manifold; appearance based object recognition; image analysis; intrinsic flows; nearest neighbor classifier; optimal linear image representation; optimization; recognition performance function; stochastic gradient algorithm; Geometry; Image analysis; Image recognition; Image storage; Image texture analysis; Nearest neighbor searches; Object recognition; Stochastic processes; Testing; Vectors; Algorithms; Artificial Intelligence; Cluster Analysis; Computer Simulation; Face; Humans; Image Enhancement; Image Interpretation, Computer-Assisted; Imaging, Three-Dimensional; Information Storage and Retrieval; Linear Models; Models, Biological; Numerical Analysis, Computer-Assisted; Pattern Recognition, Automated; Reproducibility of Results; Sensitivity and Specificity; Signal Processing, Computer-Assisted; Stochastic Processes; Subtraction Technique;

fLanguage

English

Journal_Title

Pattern Analysis and Machine Intelligence, IEEE Transactions on

Publisher

ieee

ISSN

0162-8828

Type

jour

DOI

10.1109/TPAMI.2004.1273986

Filename

1273986