SUB-CLASSIFICATION OF FARMLAND IN HIGH RESOLUTION RS IMAGES BASED ON TEXTURAL AND SPECTRAL FEATURES

摘要

The textural features are adopted to classify the farmland in high resolution remote sensing images, in which the spectral feature is similar while the textural feature is dissimilar. On the basis of analysis of the contributions to the classification caused by different textural features, the method of classification with weighted textural features is proposed. And then, the algorithm of sub-classification of farmland in high resolution remote sensing images is discussed. Finally, the experiment is given through the Brodatz combined texture image and a high resolution remote sensing image. The results show that this algorithm is effective.