The Comparison of Different Methods of Texture Analysis for Their Effcacy for Land Use Classification in Satellite Imagery
AbstractThe paper presents a comparison of the eﬃcacy of several texture analysis methods as tools for improving land use/cover classiﬁcation in satellite imagery. The tested methods were: gray level co-occurrence matrix (GLCM) features, Laplace ﬁlters and granulometric analysis, based on mathematical morphology. The performed tests included an assessment of the classiﬁcation accuracy performed based on spectro-textural datasets: spectral images with the addition of images generated using diﬀerent texture analysis methods. The class nomenclature was based on spectral and textural diﬀerences and included the following classes: water, low vegetation, bare soil, urban, and two (coniferous and deciduous) forest classes. The classiﬁcation accuracy was assessed using the overall accuracy and kappa index of agreement, based on the reference data generated using visual interpretation of the images. The analysis was performed using very high-resolution imagery (Pleiades, WorldView-2) and high-resolution imagery (Sentinel-2). The results show the eﬃcacy of selected GLCM features and granulometric analysis as tools for providing textural data, which could be used in the process of land use/cover classiﬁcation. It is also clear that texture analysis is generally a more important and eﬀective component of classiﬁcation for images of higher resolution. In addition, for classiﬁcation using GLCM results, the Random Forest variable importance analysis was performed.
|Journal series||Remote Sensing, ISSN 2072-4292|
|Publication size in sheets||0.95|
|Keywords in English||satellite imagery; classiﬁcation; texture analysis; GLCM; mathematical morphology; granulometric analysis; Laplace ﬁlter|
|ASJC Classification||; ; ; ; ;|
|Score||= 100.0, 16-06-2020, ArticleFromJournal|
|Publication indicators||= 1; = 0; : 2017 = 1.559; : 2018 = 4.118 (2) - 2018=4.74 (5)|
* presented citation count is obtained through Internet information analysis and it is close to the number calculated by the Publish or Perish system.