2010 | OriginalPaper | Buchkapitel
Retrieving Images of Similar Geometrical Configuration
verfasst von : Xiaolong Zhang, Baoxin Li
Erschienen in: Advances in Visual Computing
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Content Based Image Retrieval (CBIR) has been an active research field for a long time. Existing CBIR approaches are mostly based on low- to middle-level visual cues such as color or color histograms and possibly semantic relations of image regions, etc. In many applications, it may be of interest to retrieve images of similar geometrical configurations such as all images of a hallway-like view. In this paper we present our work on addressing such a task that seemingly requires 3D reconstruction from a single image. Our approach avoids explicit 3D reconstruction, which remains to be a challenge, through coding the potential relationship between the 3D structure of an image and its low-level features via a grid-based representation. We experimented with a data set of several thousands of images and obtained promising results.