EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian Function

We propose a two-part local image descriptor EL (Edges and Lines), based on the strongest image responses to the first- and second-order partial derivatives of the two-dimensional Gaussian function. Using the steering theorems, the proposed method finds the filter orientations giving the strongest i...

Full description

Bibliographic Details
Main Authors: Jasna Maver, Danijel Skočaj
Format: Article
Language:English
Published: Hindawi Limited 2019-01-01
Series:Mathematical Problems in Engineering
Online Access:http://dx.doi.org/10.1155/2019/1247925
id doaj-8d94d5bdd4494bc0b25518c2b72555eb
record_format Article
spelling doaj-8d94d5bdd4494bc0b25518c2b72555eb2020-11-24T22:07:22ZengHindawi LimitedMathematical Problems in Engineering1024-123X1563-51472019-01-01201910.1155/2019/12479251247925EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian FunctionJasna Maver0Danijel Skočaj1Faculty of Computer and Information Science, University of Ljubljana, SI-1000 Ljubljana, SloveniaFaculty of Computer and Information Science, University of Ljubljana, SI-1000 Ljubljana, SloveniaWe propose a two-part local image descriptor EL (Edges and Lines), based on the strongest image responses to the first- and second-order partial derivatives of the two-dimensional Gaussian function. Using the steering theorems, the proposed method finds the filter orientations giving the strongest image responses. The orientations are quantized, and the magnitudes of the image responses are histogrammed. Iterative adaptive thresholding of histogram values is then applied to normalize the histogram, thereby making the descriptor robust to nonlinear illumination changes. The two-part descriptor is empirically evaluated on the HPatches benchmark for three different tasks, namely, patch verification, image matching, and patch retrieval. The proposed EL descriptor outperforms the traditional descriptors such as SIFT and RootSIFT on all three evaluation tasks and the deep-learning-based descriptors DeepCompare, DeepDesc, and TFeat on the tasks of image matching and patch retrieval.http://dx.doi.org/10.1155/2019/1247925
collection DOAJ
language English
format Article
sources DOAJ
author Jasna Maver
Danijel Skočaj
spellingShingle Jasna Maver
Danijel Skočaj
EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian Function
Mathematical Problems in Engineering
author_facet Jasna Maver
Danijel Skočaj
author_sort Jasna Maver
title EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian Function
title_short EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian Function
title_full EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian Function
title_fullStr EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian Function
title_full_unstemmed EL: Local Image Descriptor Based on Extreme Responses to Partial Derivatives of 2D Gaussian Function
title_sort el: local image descriptor based on extreme responses to partial derivatives of 2d gaussian function
publisher Hindawi Limited
series Mathematical Problems in Engineering
issn 1024-123X
1563-5147
publishDate 2019-01-01
description We propose a two-part local image descriptor EL (Edges and Lines), based on the strongest image responses to the first- and second-order partial derivatives of the two-dimensional Gaussian function. Using the steering theorems, the proposed method finds the filter orientations giving the strongest image responses. The orientations are quantized, and the magnitudes of the image responses are histogrammed. Iterative adaptive thresholding of histogram values is then applied to normalize the histogram, thereby making the descriptor robust to nonlinear illumination changes. The two-part descriptor is empirically evaluated on the HPatches benchmark for three different tasks, namely, patch verification, image matching, and patch retrieval. The proposed EL descriptor outperforms the traditional descriptors such as SIFT and RootSIFT on all three evaluation tasks and the deep-learning-based descriptors DeepCompare, DeepDesc, and TFeat on the tasks of image matching and patch retrieval.
url http://dx.doi.org/10.1155/2019/1247925
work_keys_str_mv AT jasnamaver ellocalimagedescriptorbasedonextremeresponsestopartialderivativesof2dgaussianfunction
AT danijelskocaj ellocalimagedescriptorbasedonextremeresponsestopartialderivativesof2dgaussianfunction
_version_ 1725820774133006336