Assessment of accuracy : systematic reduction of training points for maximum likelihood classification and mixture discriminant analysis (Gaussian and t-distribution)

Simple item page
dc.contributor.authorRitchie, Michaelaen_ZA
dc.contributor.authorDebba, Praveshen_ZA
dc.contributor.authorLuck-Vogel, Melanieen_ZA
dc.contributor.authorGoodall, Victoriaen_ZA
dc.date.accessioned2021-07-10T14:31:41Z
dc.date.available2021-07-10T14:31:41Z
dc.date.issued2018-09
dc.descriptionThe original publication is available at http://sajg.org.zaen_ZA
dc.descriptionCITATION: Ritchie, M. et al. 2018. Assessment of accuracy : systematic reduction of training points for maximum likelihood classification and mixture discriminant analysis (Gaussian and t-distribution). South African Journal of Geomatics, 7(2):132-146, doi:10.4314/sajg.v7i2.3.en_ZA
dc.description.abstractRemote sensing provides a valuable tool for monitoring land cover across large areas of land. A simple yet popular method for land cover classification is Maximum Likelihood Classification (MLC), which assumes a single normal distribution of the samples per class in the feature space. Mixture Discriminant Analysis (MDA) is a natural extension of MLC which can be used with varying distributions and multiple distributions per class, which simplifies the classification process tremendously. We compare the accuracies of MLC and MDA (using a Gaussian and t-distribution) as the number of training points are systematically reduced in order to simulate varying reference data availability conditions. The results show that the more robust t-distribution MDA performs comparatively with the Gaussian MDA and that both outperform MLC when sufficient training points are available. As the number of training points increases the MDA accuracies increase while the MLC accuracy stagnates. At very low numbers of training samples (ranging from 22 to 169 dependent on the class), there is more variability in terms of which method performs best.en_ZA
dc.description.urihttp://sajg.org.za/index.php/sajg/article/view/643
dc.description.versionPublisher's version
dc.format.extent15 pages : illustrationsen_ZA
dc.identifier.citationRitchie, M. et al. 2018. Assessment of accuracy : systematic reduction of training points for maximum likelihood classification and mixture discriminant analysis (Gaussian and t-distribution). South African Journal of Geomatics, 7(2):132-146, doi:10.4314/sajg.v7i2.3.
dc.identifier.issn2225-8531 (online)
dc.identifier.otherdoi:10.4314/sajg.v7i2.3
dc.identifier.urihttp://hdl.handle.net/10019.1/110698
dc.language.isoen_ZAen_ZA
dc.publisherCONSAS Conference
dc.rights.holderAuthors retain copyright
dc.subjectRemote sensingen_ZA
dc.subjectImage processing -- Digital techniquesen_ZA
dc.subjectGauge theory (Physics)en_ZA
dc.subjectGaussian processesen_ZA
dc.subjectMachine learning -- Mathematical modelsen_ZA
dc.titleAssessment of accuracy : systematic reduction of training points for maximum likelihood classification and mixture discriminant analysis (Gaussian and t-distribution)en_ZA
dc.typeArticleen_ZA
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
ritchie_assessment_2018.pdf
Size:
703.95 KB
Format:
Adobe Portable Document Format
Description:
Download article
Download
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description:
Download
Collections
Research Articles (Geography and Environmental Studies)