MESMA与面向对象组合的土地利用分类方法

混合像元是制约传统组合分类方法精度提高的主要因素之一.为此,文章提出一种基于多端元混合像元分解(multiple endmember spectral mixture analysis,MESMA)与面向对象分类组合的分类方法,利用混合像元分解提高分类精度,借助组合方法降低"椒盐"现象影响.首先,使用M ESM A技术...

A Land Use Classification Method Based on MESMA and Object-oriented Technique

Mixed pixels are one of the main factors restricting the accuracy of traditional combined classification methods.The traditional pixel-based and object-oriented combination classification method cause inaccurate classification of ground features due to less attention to mixed pixels.To bridge this research gap,this paper proposes a new combination classification method based on the multiple endmember spectral mixture analysis(MESMA)technology and object-oriented classification.Firstly,MESMA technology is used to decompose the mixed pixels and extract the abundance information.Then,the abundance information is included in the feature variables for random forest classification based on pixel and object oriented.Finally,object-oriented and pixel-based classification results are extracted and merged.Landsat-8 image is used as the classification data source to study the land use classification of the Nalinhe No.2 mining area in Wushen Banner,Ordos City,Inner Mongolia.The results show that the classification accuracy of the proposed method is better than that of the single pixel-based,object-oriented classification method and the existing combination classification methods.By solving the mixed pixel problem,the combination classification method proposed in this paper improves the classification accuracy by 4.56%,5.66%and 4.05%,respectively,compared with pixel-based,object-oriented and current combined classification methods.MESMA technology is helpful for the combination classification method to solve the mixed pixel problem in remote sensing images with medium spatial resolution.The application of this method to remote sensing image classification can provide theoretical support for large-scale land use change monitoring and other work.