Extracting Floor Area Ratio of the Classified Buildings from Very High Resolution Satellite Image Using Multiple Features

With the rapid development of urbanization, problems such as shortage of land resource and low efficiency of land use have emerge and grown increasingly serious. The extraction of floor area ratio of the classified buildings in urban areas is of great significance to the floor area ratio(FAR) management and regulatory detailed planning in urban land development. Moreover, high-resolution data which provide analysis results of high precision are under comprehensive application in various lines of work, especially in the field of territorial resources. Given these, an object-based method for FAR extraction from very high resolution satellite image using multiple features, which are principal components, main direction, border index and rectangular fit, and Bayesian classifier is proposed. Meanwhile, FAR extraction results calculated via shadow area and shadow length are compared. We applied this method to a clip of WorldView-3 image and validated the FAR results of building units included one by one. Experimental results show that the average accuracy for shadow area method is 93.90% and 85.19% for shadow length method and shadow area method is more effective than shadow length method.