Comparison of Feature Reduction Algorithms for Classifying Tree Species With Hyperspectral Data on Three Central European Test Sites

Fassnacht, Fabian Ewald; Neumann, Christoph; Förster, Michael; Buddenbaum, Henning; Ghosh, Aniruddha; Clasen, Anne; Joshi, P. K.; Koch, Bárbara

doi:10.1109/jstars.2014.2329390

Cited by 154 publications

(118 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The classification process was conducted in ENVI 5.1 and the default parameters of classifier in the software were used. Previous studies (Fabian E. Fassnacht et al, 2014;Fung et al, 1997;Raczko & Zagajewski, 2017) suggested that about 15-20 input bands produced the best classification result. Spectral subsets comprising the first 15-20 most importance bands extracted from SDA, SVM and MNF feature selection and extraction results were served as input.…”

Section: Classificationmentioning

confidence: 92%

“…Feature selection methods select a subset from the original bands. Common selection methods include stepwise discriminate analysis, hierarchical clustering, support vector machine (SVM), partial least square discriminate analysis (PLSDA) and genetic algorithms (Fabian E. Fassnacht et al, 2014;Fung, Fung, Ma, & Siu, 1997;Pal & Foody, 2010). Feature extraction methods reduce the number of band data through data transformation with the aim to extract maximal information from the original data.…”

Section: Introductionmentioning

confidence: 99%

“…Minimum noise fraction (MNF) transformation and principal component analysis (PCA) are widely applied in hyperspectral image (Burai, Deá k, Valkó, & Tomor, 2015;Marcinkowska-Ochtyra et al, 2017). MNF is reported yielding higher accuracy than feature selection result or full band used result (Fabian E. Fassnacht et al, 2014).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Assessing the Utility of Uav-Borne Hyperspectral Image and Photogrammetry Derived 3d Data for Wetland Species Distribution Quick Mapping

Wong

Fung

2017

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

View full text Add to dashboard Cite

ABSTRACT:Lightweight unmanned aerial vehicle (UAV) loaded with novel sensors offers a low cost and minimum risk solution for data acquisition in complex environment. This study assessed the performance of UAV-based hyperspectral image and digital surface model (DSM) derived from photogrammetric point clouds for 13 species classification in wetland area of Hong Kong. Multiple feature reduction methods and different classifiers were compared. The best result was obtained when transformed components from minimum noise fraction (MNF) and DSM were combined in support vector machine (SVM) classifier. Wavelength regions at chlorophyll absorption green peak, red, red edge and Oxygen absorption at near infrared were identified for better species discrimination. In addition, input of DSM data reduces overestimation of low plant species and misclassification due to the shadow effect and inter-species morphological variation. This study establishes a framework for quick survey and update on wetland environment using UAV system. The findings indicate that the utility of UAV-borne hyperspectral and derived tree height information provides a solid foundation for further researches such as biological invasion monitoring and bio-parameters modelling in wetland.

show abstract

Section: Classificationmentioning

confidence: 92%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Assessing the Utility of Uav-Borne Hyperspectral Image and Photogrammetry Derived 3d Data for Wetland Species Distribution Quick Mapping

Wong

Fung

2017

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

View full text Add to dashboard Cite

show abstract

“…The method is widely used in areas such as hyperspectral information analysis [54][55][56]. In these areas, the dimensionality is always higher than number of observations, reaching tens and even hundreds of times the number of observations.…”

Section: Comparison With Other Low Sample Size Studiesmentioning

confidence: 99%

Feature Selection Solution with High Dimensionality and Low-Sample Size for Land Cover Classification in Object-Based Image Analysis

et al. 2017

View full text Add to dashboard Cite

Land cover information extraction through object-based image analysis (OBIA) has become an important trend in remote sensing, thanks to the increasing availability of high-resolution imagery. Segmented objects have a large number of features that cause high-dimension and low-sample size problems in the classification process. In this study, on the basis of a partial least squares generalized linear regression (PLSGLR), we propose a group corrected PLSGLR, known as G-PLSGLR, that aims to reduce the redundancy of object features for land cover identifications. Using Gaofen-2 images, the area of interest was segmented and sampled to generate small sample-size training datasets with 51 object features. The features selected by G-PLSGLR were compared against a guided regularized random forest (GRRF) in metrics of reduction rate, feature redundancy, and accuracy assessment of classification. Three indicators of overall accuracy (OA), user's accuracy (UA), and producer's accuracy (PA) were applied for accuracy assessment in this paper. The result shows that the G-PLSGLR achieved a reduction rate of 9.27 with a feature redundancy of 0.29, and a value of OA 90.63%. The GRRF achieved a reduction rate of 1.61 with a feature redundancy of 0.42, and a value of OA 85.56%. The PA of each land cover category was more than 95% using features selected by G-PLSGLR, while the PA ranged from 77 to 96% using features selected by GRRF. The UA of G-PLSGLR-selected features ranged from 70 to 80% except for grass land and bare land, which achieved 10% higher UA than GRRF-selected features. The G-PLSGLR method we proposed has the advantages of a large reduction rate, low feature redundancy, and high classification performance, which can be applied in OBIA-based land cover classification.

show abstract

“…The result shows that this methodology significantly reduces time and effort spent in executing thousands of test cases. On the other hand, some other studies use different classifiers [5], [16], [17], [18] in different domains using tools such as WEKA, Rapid Miner and ORANGE. Lee and Chan [19] proposed an approach that can be used to enhance the process of automated test tool model using machine learning technique from association rule mining.…”

Section: Data Miningmentioning

confidence: 99%

Test Case Reduction Using Data Mining Classifier Techniques

Saifan¹

2016

JSW

View full text Add to dashboard Cite

Abstract:The main purpose of test case reduction is to decrease the number of test cases in order to minimize the time and cost of executing them. We used the data mining approach, mainly because of its ability to extract patterns of test cases that are invisible. In this paper, we use two data mining classifiers, Naï ve based and J48, in order to classify a set of test cases and identify which test case is redundant or irredundant. The results show the applicability of data mining classification in removing the number of redundant test cases.

show abstract

Comparison of Feature Reduction Algorithms for Classifying Tree Species With Hyperspectral Data on Three Central European Test Sites

Cited by 154 publications

References 69 publications

Assessing the Utility of Uav-Borne Hyperspectral Image and Photogrammetry Derived 3d Data for Wetland Species Distribution Quick Mapping

Assessing the Utility of Uav-Borne Hyperspectral Image and Photogrammetry Derived 3d Data for Wetland Species Distribution Quick Mapping

Feature Selection Solution with High Dimensionality and Low-Sample Size for Land Cover Classification in Object-Based Image Analysis

Test Case Reduction Using Data Mining Classifier Techniques

Contact Info

Product

Resources

About