Robust Recovery of Eigenimages in the Presence of Outliers and Occlusions
Aleš Leonardis
; Department for Pattern Recognition and Image Processing, Technical University Vienna, Vienna, Austria and Faculty of Computer and Information Science, University of Ljubljana, Ljubljana, Slovenia
Horst Bischof
; Department for Pattern Recognition and Image Processing, Technical University Vienna, Vienna, Austria
APA 6th Edition Leonardis, A. i Bischof, H. (1996). Robust Recovery of Eigenimages in the Presence of Outliers and Occlusions. Journal of computing and information technology, 4 (1), 25-38. Preuzeto s https://hrcak.srce.hr/150304
MLA 8th Edition Leonardis, Aleš i Horst Bischof. "Robust Recovery of Eigenimages in the Presence of Outliers and Occlusions." Journal of computing and information technology, vol. 4, br. 1, 1996, str. 25-38. https://hrcak.srce.hr/150304. Citirano 21.04.2021.
Chicago 17th Edition Leonardis, Aleš i Horst Bischof. "Robust Recovery of Eigenimages in the Presence of Outliers and Occlusions." Journal of computing and information technology 4, br. 1 (1996): 25-38. https://hrcak.srce.hr/150304
Harvard Leonardis, A., i Bischof, H. (1996). 'Robust Recovery of Eigenimages in the Presence of Outliers and Occlusions', Journal of computing and information technology, 4(1), str. 25-38. Preuzeto s: https://hrcak.srce.hr/150304 (Datum pristupa: 21.04.2021.)
Vancouver Leonardis A, Bischof H. Robust Recovery of Eigenimages in the Presence of Outliers and Occlusions. Journal of computing and information technology [Internet]. 1996 [pristupljeno 21.04.2021.];4(1):25-38. Dostupno na: https://hrcak.srce.hr/150304
IEEE A. Leonardis i H. Bischof, "Robust Recovery of Eigenimages in the Presence of Outliers and Occlusions", Journal of computing and information technology, vol.4, br. 1, str. 25-38, 1996. [Online]. Dostupno na: https://hrcak.srce.hr/150304. [Citirano: 21.04.2021.]
Sažetak The basic limitations of the current appearance-based matching methods using eigenimages are non-robust estimation of coefficients and inability to cope with problems related to occlusions and segmentation. In this paper we present a new approach which successfully solves these problems. The major novelty o f our approach lies in the way how the coefficients of the eigenimages are determined. Instead of computing the coefficients by a projection of the data onto the eigenimages, we extract them by a hypothesize-and-test paradigm using subsets of image points. Compeling hypotheses are then subject to a selection procedure based on the Minimum Description Length principle. The approach enables us not only to reject outliers and to deal with occlusions but also to simultaneously use multiple classes of eigenimages.