The Classification of Land Cover Derived from High Resolution Remote Sensing Imagery

Remote sensing imagery is an attractive source of land cover information. High resolution sensing imagery provides more land cover detail than low resolution sensing imagery. Due to more complex and noisier spectral signatures for the former, new algorithms are needed to deal with high resolution imagery. Based on an integration of spectral and spatial information, a novel classification method is presented in this paper. Taking a city and countryside union region of Qingdao as the test area, a research was conducted to extract land cover information by applying classification. Five priori defined land cover classes in the classification scheme were construction, road, grassland, farmland, and water. The results show that the classification accuracy is satisfactory by the proposed method.