Open AccessFeature extraction from high resolution satellite images using K-means and colour threshold approach
Author(s) -
Ahmed S. Elsharkawy,
Mahmoud I. Abdalla
Publication year - 2019
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/610/1/012045
Subject(s) - computer science , satellite , matlab , land cover , image (mathematics) , artificial intelligence , cover (algebra) , feature extraction , remote sensing , feature (linguistics) , pattern recognition (psychology) , segmentation , image resolution , computer vision , data mining , geography , land use , engineering , mechanical engineering , linguistics , philosophy , civil engineering , aerospace engineering , operating system
In recent year developments in satellite sensors tend to the availability of high spatial and spectral resolution images. The motivation of this research paper is to obtain maximum benefits of different bands from high resolution satellite images in order to put into practice an image processing algorithm solution for extraction and classification of land cover and manmade objects can be used by non-professionals. In this research paper a novel approach for image classification is presented by applying k-means algorithm and colour threshold approach onto high resolution World View 2 (WV2) image. K-means algorithm is applied on a reflectance image to extract land cover classes and manmade objects based on a colour-based segmentation method. The proposed technique is applied through MATLAB environment. The user is asked to select few points of the desired classes and the algorithm do the rest and produce vector layers of the selected classes. The experimental results prove the effectiveness of our framework to enhance the quality of classification in aspects of computational time and precision. The preliminary results are considered promising.