A Novel Classification Technique of Landsat-8 OLI Image-Based Data Visualization: The Application of Andrews’ Plots and Fuzzy Evidential ReasoningReport as inadecuate


A Novel Classification Technique of Landsat-8 OLI Image-Based Data Visualization: The Application of Andrews’ Plots and Fuzzy Evidential Reasoning


A Novel Classification Technique of Landsat-8 OLI Image-Based Data Visualization: The Application of Andrews’ Plots and Fuzzy Evidential Reasoning - Download this document for free, or read online. Document in PDF available to download.

1

State Key Laboratory of Remote Sensing Science, Institute of Remote Sensing and Digital Earth, University of Chinese Academy of Sciences, Beijing 100094, China

2

National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China





*

Author to whom correspondence should be addressed.



Academic Editors: Xiaofeng Li and Prasad S. Thenkabail

Abstract Andrews first proposed an equation to visualize the structures within data in 1972. Since then, this equation has been used for data transformation and visualization in a wide variety of fields. However, it has yet to be applied to satellite image data. The effect of unwanted, or impure, pixels occurring in these data varies with their distribution in the image; the effect is greater if impurity pixels are included in a classifier’s training set. Andrews’ curves enable the interpreter to select outlier or impurity data that can be grouped into a new category for classification. This study overcomes the above-mentioned problem and illustrates the novelty of applying Andrews’ plots to satellite image data, and proposes a robust method for classifying the plots that combines Dempster-Shafer theory with fuzzy set theory. In addition, we present an example, obtained from real satellite images, to demonstrate the application of the proposed classification method. The accuracy and robustness of the proposed method are investigated for different training set sizes and crop types, and are compared with the results of two traditional classification methods. We find that outlier data are easily eliminated by examining Andrews’ curves and that the proposed method significantly outperforms traditional methods when considering the classification accuracy. View Full-Text

Keywords: classification; Andrews’ plots; visualization; pattern recognition; learning system; fuzzification; Dempster-Shafer theory; data transformation; Landsat 8; remote sensing data classification; Andrews’ plots; visualization; pattern recognition; learning system; fuzzification; Dempster-Shafer theory; data transformation; Landsat 8; remote sensing data





Author: Sornkitja Boonprong 1, Chunxiang Cao 1,* , Peerapong Torteeka 2 and Wei Chen 1

Source: http://mdpi.com/



DOWNLOAD PDF




Related documents