Lucas-Kanade 20 Years On: A Unifying Framework: Part 3

DMCA

Lucas-Kanade 20 Years On: A Unifying Framework: Part 3 (2002)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Simon Baker , Ralph Gross , Iain Matthews

Venue:	International Journal of Computer Vision
Citations:	706 - 30 self

BibTeX

@ARTICLE{Baker02lucas-kanade20,
    author = {Simon Baker and Ralph Gross and Iain Matthews},
    title = {Lucas-Kanade 20 Years On: A Unifying Framework: Part 3},
    journal = {International Journal of Computer Vision},
    year = {2002},
    volume = {56},
    pages = {221--255}
}

OpenURL

Abstract

Since the Lucas-Kanade algorithm was proposed in 1981 image alignment has become one of the most widely used techniques in computer vision. Applications range from optical flow, tracking, and layered motion, to mosaic construction, medical image registration, and face coding. Numerous algorithms have been proposed and a variety of extensions have been made to the original formulation. We present an overview of image alignment, describing most of the algorithms in a consistent framework. We concentrate on the inverse compositional algorithm, an efficient algorithm that we recently proposed. We examine which of the extensions to the Lucas-Kanade algorithm can be used with the inverse compositional algorithm without any significant loss of efficiency, and which cannot. In this paper, Part 3 in a series of papers, we cover the extension of image alignment to allow linear appearance variation. We first consider linear appearance variation when the error function is the Euclidean L2 norm. We describe three different algorithms, the simultaneous, project out, and normalization inverse compositional algorithms, and empirically compare them. Afterwards we consider the combination of linear appearance variation with the robust error functions described in Part 2 of this series. We first derive robust versions of the simultaneous and normalization algorithms. Since both of these algorithms are very inefficient, as in Part 2 we derive efficient approximations based on spatial coherence. We end with an empirical evaluation of the robust algorithms.

Keyphrases

linear appearance variation image alignment lucas-kanade algorithm inverse compositional algorithm derive robust version numerous algorithm face coding efficient algorithm robust algorithm significant loss normalization algorithm euclidean l2 norm computer vision different algorithm medical image registration original formulation compositional algorithm spatial coherence efficient approximation consistent framework error function robust error function optical flow empirical evaluation

DMCA