A Bag of Strings representation for Image CategorizationReportar como inadecuado

A Bag of Strings representation for Image Categorization - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

1 M2DisCo - Geometry Processing and Constrained Optimization LIRIS - Laboratoire d-InfoRmatique en Image et Systèmes d-information 2 FT R&D - France Télécom Recherche & Développement

Abstract : This paper presents an architecture well suited for natu-ral images classification or visual object recognition applications. Themethod proposes to integrate a spatial representation into the well known”bag of local signatures” approach. For this purpose, it combines thepower of a string representation which provides an ordered view of localfeatures with the vectorial histogram representation allowing to recognizeefficiently and quickly an image by using a machine learning classifier.To reach this goal, we propose to represent an image by a set of stringsof local signatures obtained by tracking the detected salient points alongimage edges. We propose here to conjointly use the Holder exponentsand the direction of minimal regularity of the bidimensionnal signal sin-gularities to compute a signature describing precisely a region of interestcentered on an interest point. As we will see, an alphabet of strings iseasily obtained by using a typical self organizing map architecture. Asa consequence, a ”bag of strings” representation is used, providing acompact representation encoding both local signatures and spatial infor-mation. This representation is particularly well suited to train a supportvector machine classifier used for the last classification step. This archi-tecture obtains good classification rates on different well known datasets.

Autor: Julien Ros - Christophe Laurent - Jean-Michel Jolion -

Fuente: https://hal.archives-ouvertes.fr/


Documentos relacionados