We introduce a set of distortion adaptive Sobel filters for the direct estimation of geometrically correct gradients of wide angle images. The definition of the filters is based on Sobel's rationale and accounts for the geometric transformation undergone by wide angle images due to the presence of radial distortion. Moreover, we show that a local normalization of the filters magnitude is essential to achieve state-of-the-art results. To perform the experimental analysis, we propose an evaluation pipeline and a benchmark dataset of images belonging to different scene categories. Experiments on both, synthetic and real images, show that our approach outperforms the current state-of-the-art in both gradient estimation and keypoint matching for images characterized by large amounts of radial distortion. The collected dataset and the MATLAB code of the proposed method can be downloaded at our web page http://iplab.dmi.unict.it/DASF/. © 2017 Elsevier Inc.

We introduce a set of distortion adaptive Sobel filters for the direct estimation of geometrically correct gradients of wide angle images. The definition of the filters is based on Sobel's rationale and accounts for the geometric transformation undergone by wide angle images due to the presence of radial distortion. Moreover, we show that a local normalization of the filters magnitude is essential to achieve state-of-the-art results. To perform the experimental analysis, we propose an evaluation pipeline and a benchmark dataset of images belonging to different scene categories. Experiments on both, synthetic and real images, show that our approach outperforms the current state-of-the-art in both gradient estimation and keypoint matching for images characterized by large amounts of radial distortion. The collected dataset and the MATLAB code of the proposed method can be downloaded at our web page http://iplab.dmi.unict.it/DASF/

Distortion adaptive Sobel filters for the gradient estimation of wide angle images

Furnari A;FARINELLA, GIOVANNI MARIA;BATTIATO, SEBASTIANO
2017-01-01

Abstract

We introduce a set of distortion adaptive Sobel filters for the direct estimation of geometrically correct gradients of wide angle images. The definition of the filters is based on Sobel's rationale and accounts for the geometric transformation undergone by wide angle images due to the presence of radial distortion. Moreover, we show that a local normalization of the filters magnitude is essential to achieve state-of-the-art results. To perform the experimental analysis, we propose an evaluation pipeline and a benchmark dataset of images belonging to different scene categories. Experiments on both, synthetic and real images, show that our approach outperforms the current state-of-the-art in both gradient estimation and keypoint matching for images characterized by large amounts of radial distortion. The collected dataset and the MATLAB code of the proposed method can be downloaded at our web page http://iplab.dmi.unict.it/DASF/. © 2017 Elsevier Inc.
2017
We introduce a set of distortion adaptive Sobel filters for the direct estimation of geometrically correct gradients of wide angle images. The definition of the filters is based on Sobel's rationale and accounts for the geometric transformation undergone by wide angle images due to the presence of radial distortion. Moreover, we show that a local normalization of the filters magnitude is essential to achieve state-of-the-art results. To perform the experimental analysis, we propose an evaluation pipeline and a benchmark dataset of images belonging to different scene categories. Experiments on both, synthetic and real images, show that our approach outperforms the current state-of-the-art in both gradient estimation and keypoint matching for images characterized by large amounts of radial distortion. The collected dataset and the MATLAB code of the proposed method can be downloaded at our web page http://iplab.dmi.unict.it/DASF/
Adaptive filters; Gradient estimation; Wide angle images
File in questo prodotto:
File Dimensione Formato  
Distortion adaptive Sobel filters.pdf

solo gestori archivio

Tipologia: Versione Editoriale (PDF)
Licenza: Creative commons
Dimensione 4.18 MB
Formato Adobe PDF
4.18 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.11769/21885
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 22
  • ???jsp.display-item.citation.isi??? 19
social impact