Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical Flow
In two-color multiview (2CMV) advanced geospatial information (AGI) products, temporal changes in synthetic aperture radar (SAR) images acquired at different times are detected, colorized, and overlaid on an initial image such that new features are represented in cyan, and features that have disappe...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2019-06-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/19/11/2605 |
_version_ | 1798003444862156800 |
---|---|
author | Berkay Kanberoglu David Frakes |
author_facet | Berkay Kanberoglu David Frakes |
author_sort | Berkay Kanberoglu |
collection | DOAJ |
description | In two-color multiview (2CMV) advanced geospatial information (AGI) products, temporal changes in synthetic aperture radar (SAR) images acquired at different times are detected, colorized, and overlaid on an initial image such that new features are represented in cyan, and features that have disappeared are represented in red. Accurate detection of temporal changes in 2CMV AGI products can be challenging because of ’speckle noise’ susceptibility and false positives that result from small orientation differences between objects imaged at different times. Accordingly, 2CMV products are often dominated by colored pixels when changes are detected via simple pixel-wise cross-correlation. The state-of-the-art in SAR image processing demonstrates that generating efficient 2CMV products, while accounting for the aforementioned problem cases, has not been well addressed. We propose a methodology to address the aforementioned two problem cases. Before detecting temporal changes, speckle and smoothing filters mitigate the effects of speckle noise. To detect temporal changes, we propose using unsupervised feature learning algorithms in conjunction with optical flow algorithms that track the motion of objects across time in small regions of interest. The proposed framework for distinguishing between actual motion and misregistration can lead to more accurate and meaningful change detection and improve object extraction from an SAR AGI product. |
first_indexed | 2024-04-11T12:07:50Z |
format | Article |
id | doaj.art-cb608dcfb50a45d882596c58d234eab3 |
institution | Directory Open Access Journal |
issn | 1424-8220 |
language | English |
last_indexed | 2024-04-11T12:07:50Z |
publishDate | 2019-06-01 |
publisher | MDPI AG |
record_format | Article |
series | Sensors |
spelling | doaj.art-cb608dcfb50a45d882596c58d234eab32022-12-22T04:24:41ZengMDPI AGSensors1424-82202019-06-011911260510.3390/s19112605s19112605Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical FlowBerkay Kanberoglu0David Frakes1School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85281, USASchool of Biological and Health Systems Engineering, Arizona State University, Tempe, AZ 85281, USAIn two-color multiview (2CMV) advanced geospatial information (AGI) products, temporal changes in synthetic aperture radar (SAR) images acquired at different times are detected, colorized, and overlaid on an initial image such that new features are represented in cyan, and features that have disappeared are represented in red. Accurate detection of temporal changes in 2CMV AGI products can be challenging because of ’speckle noise’ susceptibility and false positives that result from small orientation differences between objects imaged at different times. Accordingly, 2CMV products are often dominated by colored pixels when changes are detected via simple pixel-wise cross-correlation. The state-of-the-art in SAR image processing demonstrates that generating efficient 2CMV products, while accounting for the aforementioned problem cases, has not been well addressed. We propose a methodology to address the aforementioned two problem cases. Before detecting temporal changes, speckle and smoothing filters mitigate the effects of speckle noise. To detect temporal changes, we propose using unsupervised feature learning algorithms in conjunction with optical flow algorithms that track the motion of objects across time in small regions of interest. The proposed framework for distinguishing between actual motion and misregistration can lead to more accurate and meaningful change detection and improve object extraction from an SAR AGI product.https://www.mdpi.com/1424-8220/19/11/2605SAR2CMVchange detectionoptical flowk-meansK-SVD |
spellingShingle | Berkay Kanberoglu David Frakes Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical Flow Sensors SAR 2CMV change detection optical flow k-means K-SVD |
title | Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical Flow |
title_full | Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical Flow |
title_fullStr | Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical Flow |
title_full_unstemmed | Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical Flow |
title_short | Improving the Accuracy of Two-Color Multiview (2CMV) Advanced Geospatial Information (AGI) Products Using Unsupervised Feature Learning and Optical Flow |
title_sort | improving the accuracy of two color multiview 2cmv advanced geospatial information agi products using unsupervised feature learning and optical flow |
topic | SAR 2CMV change detection optical flow k-means K-SVD |
url | https://www.mdpi.com/1424-8220/19/11/2605 |
work_keys_str_mv | AT berkaykanberoglu improvingtheaccuracyoftwocolormultiview2cmvadvancedgeospatialinformationagiproductsusingunsupervisedfeaturelearningandopticalflow AT davidfrakes improvingtheaccuracyoftwocolormultiview2cmvadvancedgeospatialinformationagiproductsusingunsupervisedfeaturelearningandopticalflow |