α-MeanShift++: Improving MeanShift++ for Image Segmentation
MeanShift is one of the popular clustering algorithms and can be used to partition a digital image into semantically meaningful regions in an unsupervised manner. However, due to its prohibitively high computational complexity, a grid-based approach, called MeanShift++, has rec...
Main Author: | |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2021-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9541345/ |
id |
doaj-a5eea64dbc5441d495fbe07abf13e928 |
---|---|
record_format |
Article |
spelling |
doaj-a5eea64dbc5441d495fbe07abf13e9282021-09-30T23:00:39ZengIEEEIEEE Access2169-35362021-01-01913143013143910.1109/ACCESS.2021.31142239541345α-MeanShift++: Improving MeanShift++ for Image SegmentationHanhoon Park0https://orcid.org/0000-0002-6968-4565Department of Electronic Engineering, Pukyong National University, Busan, South KoreaMeanShift is one of the popular clustering algorithms and can be used to partition a digital image into semantically meaningful regions in an unsupervised manner. However, due to its prohibitively high computational complexity, a grid-based approach, called MeanShift++, has recently been proposed and succeeded to surprisingly reduce the computational complexity of MeanShift. Nevertheless, we found that MeanShift++ still has computational redundancy and there is room for improvement in terms of accuracy and runtime; thus, we propose an improvement to MeanShift++, named <inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>-MeanShift++. We first attempt to minimize the computational redundancy by using an additional hash table. Then, we introduce a speedup factor (<inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>) to reduce the number of iterations required until convergence, and we use more neighboring grid cells for the same bandwidth to improve accuracy. Through intensive experiments on image segmentation benchmark datasets, we demonstrate that <inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>-MeanShift++ can run 4.1-<inline-formula> <tex-math notation="LaTeX">$4.6\times $ </tex-math></inline-formula> faster on average (but up to <inline-formula> <tex-math notation="LaTeX">$7\times $ </tex-math></inline-formula>) than MeanShift++ and achieve better image segmentation quality.https://ieeexplore.ieee.org/document/9541345/Clusteringmean shift algorithmMeanShift++image segmentation |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Hanhoon Park |
spellingShingle |
Hanhoon Park α-MeanShift++: Improving MeanShift++ for Image Segmentation IEEE Access Clustering mean shift algorithm MeanShift++ image segmentation |
author_facet |
Hanhoon Park |
author_sort |
Hanhoon Park |
title |
α-MeanShift++: Improving MeanShift++ for Image Segmentation |
title_short |
α-MeanShift++: Improving MeanShift++ for Image Segmentation |
title_full |
α-MeanShift++: Improving MeanShift++ for Image Segmentation |
title_fullStr |
α-MeanShift++: Improving MeanShift++ for Image Segmentation |
title_full_unstemmed |
α-MeanShift++: Improving MeanShift++ for Image Segmentation |
title_sort |
α-meanshift++: improving meanshift++ for image segmentation |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2021-01-01 |
description |
MeanShift is one of the popular clustering algorithms and can be used to partition a digital image into semantically meaningful regions in an unsupervised manner. However, due to its prohibitively high computational complexity, a grid-based approach, called MeanShift++, has recently been proposed and succeeded to surprisingly reduce the computational complexity of MeanShift. Nevertheless, we found that MeanShift++ still has computational redundancy and there is room for improvement in terms of accuracy and runtime; thus, we propose an improvement to MeanShift++, named <inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>-MeanShift++. We first attempt to minimize the computational redundancy by using an additional hash table. Then, we introduce a speedup factor (<inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>) to reduce the number of iterations required until convergence, and we use more neighboring grid cells for the same bandwidth to improve accuracy. Through intensive experiments on image segmentation benchmark datasets, we demonstrate that <inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>-MeanShift++ can run 4.1-<inline-formula> <tex-math notation="LaTeX">$4.6\times $ </tex-math></inline-formula> faster on average (but up to <inline-formula> <tex-math notation="LaTeX">$7\times $ </tex-math></inline-formula>) than MeanShift++ and achieve better image segmentation quality. |
topic |
Clustering mean shift algorithm MeanShift++ image segmentation |
url |
https://ieeexplore.ieee.org/document/9541345/ |
work_keys_str_mv |
AT hanhoonpark x03b1meanshiftx002bx002bimprovingmeanshiftx002bx002bforimagesegmentation |
_version_ |
1716862646147350528 |