Classification in High-Dimensional Feature Spaces—Assessment Using SVM, IVM and RVM With Focus on Simulated EnMAP Data

Braun, Andreas; Weidner, Uwe; Hinz, Stefan

doi:10.1109/jstars.2012.2190266

Cited by 54 publications

(33 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore the implemented version of IVM utilised here is a hybrid forward/backward strategy that adds import vectors whilst concurrently testing if they can be removed in each step, thus leading to a sparse and more accurate solution [41]. Furthermore, the IVM selects data points from the entire distribution resulting in a smoother decision boundary which is based on the optimal separating hyperplane in multidimensional space compared to that of SVM algorithms [42]. The benefits of the IVM algorithm have resulted in this approach being successfully applied in a number of studies (e.g., [42][43][44][45]) due to its accuracy and performance advantages over alternative methodologies including SVM and the traditional Maximum Likelihood (ML) classifiers [44,45].…”

Section: Data Classificationmentioning

confidence: 99%

Urban Growth Dynamics in Perth, Western Australia: Using Applied Remote Sensing for Sustainable Future Planning

MacLachlan

Biggs

Roberts

et al. 2017

Land

View full text Add to dashboard Cite

Earth observation data can provide valuable assessments for monitoring the spatial extent of (un)sustainable urban growth of the world's cities to better inform planning policy in reducing associated economic, social and environmental costs. Western Australia has witnessed rapid economic expansion since the turn of the century founded upon extensive natural resource extraction. Thus, Perth, the state capital of Western Australia, has encountered significant population and urban growth in response to the booming state economy. However, the recent economic slowdown resulted in the largest decrease in natural resource values that Western Australia has ever experienced. Here, we present multi-temporal urban expansion statistics from 1990 to 2015 for Perth, derived from Landsat imagery. Current urban estimates used for future development plans and progress monitoring of infill and density targets are based upon aggregated census data and metrics unrepresentative of actual land cover change, underestimating overall urban area. Earth observation provides a temporally consistent methodology, identifying areal urban area at higher spatial and temporal resolution than current estimates. Our results indicate that the spatial extent of the Perth Metropolitan Region has increased 45% between 1990 and 2015, over 320 km 2 . We highlight the applicability of earth observation data in accurately quantifying urban area for sustainable targeted planning practices.

show abstract

Section: Data Classificationmentioning

confidence: 99%

Urban Growth Dynamics in Perth, Western Australia: Using Applied Remote Sensing for Sustainable Future Planning

MacLachlan

Biggs

Roberts

et al. 2017

Land

View full text Add to dashboard Cite

show abstract

“…IVM generates two outputs: a soft (subpixel) data set that defines the probability of a pixel containing a given classification value (e.g. land-cover type) and a traditional 'hardened' classified data set (Braun, Weidner, and Hinz 2012). Training samples were collected from the 12 and 19 July 2005 Landsat 5 TM image composite, coinciding with peak vegetation greenness which provides the greatest spectral separability between vegetated and non-vegetated surfaces (Feyisa et al 2016;Chen et al 2014).…”

Section: Landsat Classificationmentioning

confidence: 99%

“…IVM has been found to consistently outperform decision trees, artificial neural networks, and maximum likelihood algorithms (Watanachaturaporn, Arora, and Varshney 2008;Kotsiantis, Zaharakis, and Pintelas 2006;Huang, Davis, and Townshend 2002), with preferential (Braun, Weidner, and Hinz 2012) and comparable results to SVM (Roscher, Waske, and Forstner 2010). However, due to the heterogeneity of urban areas, it is important to calibrate these subpixel approaches against high spatial resolution data that capture the diverse characteristics found within urban environments (Lu, Moran, and Hetrick 2011).…”

Section: Introductionmentioning

confidence: 99%

Subpixel land-cover classification for improved urban area estimates using Landsat

MacLachlan

Roberts

Biggs

et al. 2017

International Journal of Remote Sensing

View full text Add to dashboard Cite

Urban areas are Earth's fastest growing land use that impact hydrological and ecological systems and the surface energy balance. The identification and extraction of accurate spatial information relating to urban areas is essential for future sustainable city planning owing to its importance within global environmental change and human-environment interactions. However, monitoring urban expansion using medium resolution (30-250 m) imagery remains challenging due to the variety of surface materials that contribute to measured reflectance resulting in spectrally mixed pixels. This research integrates high spatial resolution orthophotos and Landsat imagery to identify differences across a range of diverse urban subsets within the rapidly expanding Perth Metropolitan Region (PMR), Western Australia. Results indicate that calibrating Landsat-derived subpixel land-cover estimates with correction values (calculated from spatially explicit comparisons of subpixel Landsat values to classified high-resolution data which accounts for over [under] estimations of Landsat) reduces moderate resolution urban area over (under) estimates by on an average 55.08% for the PMR. This approach can be applied to other urban areas globally through use of frequently available and/or low-cost high spatial resolution imagery (e.g. using Google Earth). This will improve urban growth estimations to help monitor and measure change whilst providing metrics to facilitate sustainable urban development targets within cities around the world. ARTICLE HISTORY

show abstract

“…In addition, the significance of differences among these classification algorithms was estimated using the McNemar's significance testing. The McNemar's test based on a chi-square statistics was done using the following definition [52,53]:…”

Section: Accuracy Assessment and Error And Uncertainty Analysismentioning

confidence: 99%

Mapping Oil Palm Plantations in Cameroon Using PALSAR 50-m Orthorectified Mosaic Images

Dong

Tenku

et al. 2015

Remote Sensing

View full text Add to dashboard Cite

Oil palm plantations have expanded rapidly. Estimating either positive effects on the economy, or negative effects on the environment, requires accurate maps. In this paper, three classification algorithms (Support Vector Machine (SVM), Decision Tree and K-Means) were explored to map oil palm plantations in Cameroon, using PALSAR 50 m Orthorectified Mosaic images and differently sized training samples. SVM had the ideal performance with overall accuracy ranging from 86% to 92% and a Kappa coefficient from 0.76 to 0.85, depending upon the training sample size (ranging from 20 to 500 pixels per class). The advantage of SVM was more obvious when the training sample size was smaller. K-Means required the user's intervention, and thus, the accuracy depended on the level of his/her expertise and experience. For large-scale mapping of oil palm plantations, the Decision Tree algorithm outperformed both SVM and K-Means in terms of speed and performance. In addition, the decision threshold values of Decision Tree for a large training sample size agrees with the results from previous studies, which implies the possible universality of the decision threshold. If it can be verified, the Decision Tree algorithm will be an easy and robust methodology for mapping oil palm plantations. OPEN ACCESSRemote Sens. 2015, 7 1207

show abstract

Classification in High-Dimensional Feature Spaces—Assessment Using SVM, IVM and RVM With Focus on Simulated EnMAP Data

Cited by 54 publications

References 14 publications

Urban Growth Dynamics in Perth, Western Australia: Using Applied Remote Sensing for Sustainable Future Planning

Urban Growth Dynamics in Perth, Western Australia: Using Applied Remote Sensing for Sustainable Future Planning

Subpixel land-cover classification for improved urban area estimates using Landsat

Mapping Oil Palm Plantations in Cameroon Using PALSAR 50-m Orthorectified Mosaic Images

Contact Info

Product

Resources

About