Comparison of object-oriented and Maximum Likelihood Classification of land use in Karst area

Abstract

This paper presents a comparison analysis for methods of land use classification, between object-oriented and Maximum likelihood methods in Karst area. Nine types of objects are selected in terms of “Standard of Land Classification” (version 2001)” and “Classification of Land Use (version 2008)”, China for classification. They are Shrub land, dry land, construction land, bare land, woodland, irrigation land, water, forest land, and garden land. Comparison analysis of land uses using object-oriented and Maximum Likelihood Classification method is conducted. Downtown of Daxu county, Guilin, China in a typical karst area is chosen as the study area, and ISR-P6 satellite imagery are chosen for experiments. The experimental results discover that classification results have a significant impacts to land use classification in karst area. The object-oriented classifier achieved the classification accuracy of 93.96%, whereas the maximum likelihood classifier produced a classification accuracy of 78.94%.This study demonstrates that the object-oriented classifier is significantly better for classification of land use in karst area.