Modified ALV for selecting the optimal spatial resolution and its scale effect on image classification accuracy

  • Authors:

  • Dongping Ming;Jianyu Yang;Longxiang Li;Zhuoqin Song

  • Affiliations:

  • School of Information Engineering, China University of Geosciences (Beijing), Xueyuan Road 29, Haidian, Beijing, 100083, China and The State Key Laboratory of Remote Sensing Science, Institute of ...;College of Information & Electronic Engineering, China Agricultural University, Tsinghua East Road 17, Haidian, Beijing, 100083, China;School of Information Engineering, China University of Geosciences (Beijing), Xueyuan Road 29, Haidian, Beijing, 100083, China;School of Information Engineering, China University of Geosciences (Beijing), Xueyuan Road 29, Haidian, Beijing, 100083, China

  • Venue:
  • Mathematical and Computer Modelling: An International Journal
  • Year:
  • 2011

Quantified Score

Hi-index 0.98

Visualization

Abstract

Faced with the prevalence of multi-spatial resolution satellite data sets, selecting data with an appropriate resolution has become a new problem. This paper analyses the significance of scale selection of remote sensing images and discusses a geo-statistics based method for selecting the optimal spatial resolution of a remote sensing image. Breaking through the limitation of traditional average local variance (ALV), the modified ALV method based on variable window size and variable resolution is proposed to quantitatively select the optimal spatial resolution of a remote sensing image. In order to verify the validity of this method and interpret the relationship among spatial resolution, local variance and classification accuracy, this paper gives further image classification experiments with different spatial resolution. The experimental results show that the trend of classification accuracy along spatial resolution is basically accordant with that of modified ALV, which means that the image classification accuracy of the optimal resolution image is basically higher than those of others. Therefore, the modified ALV method for quantitively selecting the optimal spatial resolution of remote sensing image has theoretical and instructional meaning to a certain extent.