An object based building extraction method and classification using high resolution remote sensing data

The increasing availability of the high spatial resolution satellite images has provided a new data source for building extraction. This paper proves the concept of object oriented classification using high-resolution satellite data (Cartosat-1 satellite data fused with IRS-1C, LISS IV data) for automatic building extraction using eCognition software. In this study, the overall accuracy of classified image is 0.94 and Kappa accuracy is 0.92. The producer accuracy for building, vegetation and shadow are 0.9745, 1.0 and 0.8999, respectively, whereas user accuracy for building, vegetation and shadow are 1.0, 0.9475 and 1.0 respectively. The classification overall accuracy is based on TTA mask (training and test area mask) and it is 0.98 and Kappa accuracy is 0.96. The producer accuracy for building, forest and shadow are 1.0, 1.0 and 0.7344, respectively and user accuracy for building, vegetation and shadow are 1.0, 0.9475 and 1.0, respectively.