Support vector machines for classification in remote sensing

Pal, Mahesh; Mather, Paul M.

doi:10.1080/01431160512331314083

Cited by 858 publications

(423 citation statements)

References 4 publications

Supporting

Mentioning

361

Contrasting

Unclassified

Order By: Relevance

“…Figure 3 illustrates a simple scenario of a two-class separable classification problem in a two-dimensional input space [30]. SVM represent a noticeable development in machine learning research [31], particularly appealing in the remote sensing field due to is ability to generalize well, even with limited training samples, which is a common limitation for remote sensing applications [30]. It can achieve higher classification accuracy than Maximum Likelihood and Artificial Neural Network [29] with an overall accuracy above 90% [32].…”

Section: Lulc Maps Production and Lulcc Detectionmentioning

confidence: 99%

Land Cover Mapping Analysis and Urban Growth Modelling Using Remote Sensing Techniques in Greater Cairo Region—Egypt

Megahed

Cabral

Silva

et al. 2015

IJGI

169

141

View full text Add to dashboard Cite

This study modeled the urban growth in the Greater Cairo Region (GCR), one of the fastest growing mega cities in the world, using remote sensing data and ancillary data. Three land use land cover (LULC) maps (1984, 2003 and 2014) were produced from satellite images by using Support Vector Machines (SVM). Then, land cover changes were detected by applying a high level mapping technique that combines binary maps (change/no-change) and post classification comparison technique. The spatial and temporal urban growth patterns were analyzed using selected statistical metrics developed in the FRAGSTATS software. Major transitions to urban were modeled to predict the future scenarios for year 2025 using Land Change Modeler (LCM) embedded in the IDRISI software. The model results, after validation, indicated that 14% of the vegetation and 4% of the desert in 2014 will be urbanized in 2025. The urban areas within a 5-km buffer around: the Great Pyramids, Islamic Cairo and Al-Baron Palace were calculated, highlighting an intense urbanization especially around the Pyramids; 28% in 2014 up to 40% in 2025. Knowing the current and estimated urbanization situation in GCR will help decision makers to adjust and develop new plans to achieve a sustainable development of urban areas and to protect the historical locations.

show abstract

Section: Lulc Maps Production and Lulcc Detectionmentioning

confidence: 99%

Land Cover Mapping Analysis and Urban Growth Modelling Using Remote Sensing Techniques in Greater Cairo Region—Egypt

Megahed

Cabral

Silva

et al. 2015

IJGI

169

141

View full text Add to dashboard Cite

show abstract

“…A report by Huang et al [2002] concluded that the accuracy improvement was insignificant if γ exceeded 7.5. Using a combined Radial Basis Function (RBF) kernel and C = 5000, Pal and Mather [2005] suggested γ = 2 to gain the best accuracy. In another case study, Foody and Mathur [2004] showed that optimal γ was in range between 0.005 and 0.08.…”

Section: Introductionmentioning

confidence: 99%

Mapping stand age of rubber plantation using ALOS-2 polarimetric SAR data

Trisasongko

2017

European Journal of Remote Sensing

View full text Add to dashboard Cite

This paper presents an evaluation on strategies for rubber plantation mapping employing SAR data coupled with Random Forest (RF) and Support Vector Machine (SVM). Linear backscatter coefficients achieved saturation point at about 10 years, making this form of polarimetric data being robust only for young to mature stands. This research found that the performance of both algorithms was comparable. The addition of texture features gave substantial impact to the overall accuracy. As indicated by the analysis of variable importance, only some texture features have contributed to higher overall accuracy. Classification using a subset of texture features pointed out that accuracy could be improved using dual polarimetric data, while trivial enhancement was seen in combined HH, HV and VV backscatter intensities. The research showed that classification accuracy could be further augmented by setting proper classification parameters. Nonetheless, it is argued that the level of improvement would greatly depend on selecting a proper dataset fed into classifier, rather than tuning classifier parameters. ARTICLE HISTORY

show abstract

“…In feature extraction, a decision plane is a plane that separates between a set of objects having different class memberships. Points closest to the margin are called support vectors and are used for the training phase of classification (Huang & Townshend, 2002;Mountrakis & Ogole, 2011;Pal & Mather, 2005). …”

Section: Figure 1 General Flowchart Of Lidar Data Classificationmentioning

confidence: 99%

Quality Assessment and Control of Outputs of a Nationwide Agricultural Land Cover Mapping Program Using Lidar: Phil-Lidar 2 Parmap Experience

Pagkalinawan

Gatdula

Jose

et al. 2017

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:The Agricultural Resources Extraction from LiDAR Surveys (PARMAP) project component of the Nationwide Detailed Resources Assessment using LiDAR (Phil-LiDAR 2) Program aims to produce detailed agricultural maps using LiDAR. Agricultural land cover at crop level was classified through object based image analysis using Support Vector Machine as classifier and LiDAR derivatives from point cloud (2 points per sq.m.) and orthophoto (0.5-meter resolution) as inputs. An accuracy of at least 90%, assessed using validation points from the field and through image interpretation, was required before proceeding to post-processing and map lay-out. Knowledge sharing and capacity development facilitated by the University of the Philippines Diliman (UPD) enabled partner universities across the Philippines to produce outputs for their assigned region. Considering output layers were generated by multiple teams working on different landscape complexities with some degree of data quality variability, quality checking is crucial to ensure accuracy standards were met. UPD PARMap devised a centralized and end-to-end scheme divided into four steps -land classification, GIS post-processing, schema application, and map lay-out. At each step, a block is reviewed and, subsequently, either approved or returned with documentation on required revisions. Turnaround time of review is at least one block (area ranging from 10 to 580 sq. km.) per day. For coastal municipalities, an additional integration process to incorporate mapped coastal features was applied. Common problems observed during quality checking include misclassifications, gaps between features, incomplete attributes and missing map elements. Some issues are particular to specific blocks such as problematic LiDAR derivatives. UPD addressed these problems through discussion and mentoring visits to partner universities. As of March 2017, a total of 336 municipal agricultural maps have been turned-over to various stakeholders. For the remaining months of the program, an additional 360 maps are expected to be distributed.

show abstract

Support vector machines for classification in remote sensing

Cited by 858 publications

References 4 publications

Land Cover Mapping Analysis and Urban Growth Modelling Using Remote Sensing Techniques in Greater Cairo Region—Egypt

Land Cover Mapping Analysis and Urban Growth Modelling Using Remote Sensing Techniques in Greater Cairo Region—Egypt

Mapping stand age of rubber plantation using ALOS-2 polarimetric SAR data

Quality Assessment and Control of Outputs of a Nationwide Agricultural Land Cover Mapping Program Using Lidar: Phil-Lidar 2 Parmap Experience

Contact Info

Product

Resources

About