Comparing Two Occluded Object Recognition Systems using Subspace Graph Search Methods

D.J. Pack and C.H.G. Wright (USA)


Recognition, Occluded, Subspace, Eigenspace


Previous work by the authors have described a method of recognizing objects in an image, even when these objects are partially occluded, using a graph search technique over a reduced-dimensional eigenvector subspace. This paper describes a comparison of two techniques, one searching the column space of the reduced-dimensional subspace and one searching the row space of the reduced-dimensional subspace. The recognition accuracy and computational complexity of the two methods are used as figures of merit.

Important Links:

Go Back