In EDS ansehen

Exploring the representation capabilities of the HOG descriptor

Gespeichert in:
Bibliographische Detailangaben
Titel: Exploring the representation capabilities of the HOG descriptor
Autoren: Tatu, Aditya Jayant, Lauze, Francois Bernard, Nielsen, Mads, Kimia, Benjamin
Quelle: 2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops). :1410-1417
Verlagsinformationen: IEEE, 2011.
Publikationsjahr: 2011
Schlagwörter: Histograms, Equations, random images, visual databases, noncategory images, 02 engineering and technology, dense grid of points, source image, initial visual appearance, object content, Mathematical model, SIFT, PASCAL databases, object recognition strategies, 0202 electrical engineering, electronic engineering, information engineering, HOG descriptor, image representation, ImageNet, LabelMe, regional descriptors, very large databases, Object recognition, Vectors, image recognition, CALTECH databases, Educational institutions, sparse set of points, metameric class, Upper bound, impostor image sharing, representation capabilities
Beschreibung: Object recognition strategies are increasingly based on regional descriptors such as SIFT or HOG at a sparse set of points or on a dense grid of points. Despite their success on databases such as PASCAL and CALTECH, the capability of such a representation in capturing the essential object content of the image is not well-understood: How large is the equivalence class of images sharing the same HOG descriptor? Are all these images from the same object category, and if not, do the non-category images resemble random images which cannot generically arise from imaged scenes? How frequently do images from two categories share the same HOG-based representation? These questions are increasingly more relevant as very large databases such as ImageNet and LabelMe are being developed where the current object recognition strategies show limited success. We examine these questions by introducing the metameric class of moments of HOG which allows for a target image to be morphed into an impostor image sharing the HOG representation of a source image while retaining the initial visual appearance. We report that two distinct images can be made to share the same HOG representation when the overlap between HOG patches is minimal, and the success of this method falls with increasing overlap. This paper is therefore a step in the direction of developing a sampling theorem for representing images by HOG features.
Publikationsart: Article
DOI: 10.1109/iccvw.2011.6130416
Dokumentencode: edsair.doi.dedup.....f70ff04b0c28f1db8f9a877fdcec2c94
Datenbank: OpenAIRE
Beschreibung
Abstract:Object recognition strategies are increasingly based on regional descriptors such as SIFT or HOG at a sparse set of points or on a dense grid of points. Despite their success on databases such as PASCAL and CALTECH, the capability of such a representation in capturing the essential object content of the image is not well-understood: How large is the equivalence class of images sharing the same HOG descriptor? Are all these images from the same object category, and if not, do the non-category images resemble random images which cannot generically arise from imaged scenes? How frequently do images from two categories share the same HOG-based representation? These questions are increasingly more relevant as very large databases such as ImageNet and LabelMe are being developed where the current object recognition strategies show limited success. We examine these questions by introducing the metameric class of moments of HOG which allows for a target image to be morphed into an impostor image sharing the HOG representation of a source image while retaining the initial visual appearance. We report that two distinct images can be made to share the same HOG representation when the overlap between HOG patches is minimal, and the success of this method falls with increasing overlap. This paper is therefore a step in the direction of developing a sampling theorem for representing images by HOG features.
DOI:10.1109/iccvw.2011.6130416