A Hybrid Approach to Improve Classification Accuracy of Mapping Perennial Crops in Bảo Lâm District, Lâm Đồng Province

Abstract

Perennial crops deliver strong economic, social and ecological benefits to many tropical countries. Accurate maps acquired through the remote sensing of the perennial crops are not yet available in the Central Highlands. The main objective of this study is to improve classification accuracy when mapping perennial crops in Bảo Lâm district with a hybrid approach and pixel-based classification. The results suggest improvements in the coffee and tea crop classification accuracy as compared to earlier work, and the overall accuracy increases from 71percent to 75percent and the Kappa coefficient improves from 0.66 to 0.71. The input parameters that most strongly impact the classification accuracy are NDVI and slope. This paper offers a comprehensive study on the accuracy gains through using a hybrid approach.