A Study Of Building Extraction From High Resolution Remote Sensing Images

Building extraction from high resolution remote sensing images is an internationally advanced research field in remote sensing, photogrammetry and computer vision. This research has significant academic and practical value. But there are many difficulties, a lot of work is needed to make it a mature technique. We made some research in this field, developed a basic building model. The corresponding solutions are given.The work including: (1) A right-angled polygon building model is introduced, which is based on rectangle structure graphics. A building extraction algorithm based on rectangle structure graphics and an algorithm based on CART is proposed from the model. Both of them can extract the outlines of complex buildings, and has high noise tolerance. (2) Researches on main direction of buildings are made. The main direction of building is an important parameter which denotes building's heading, it is also a primary step in building extraction. A lot of researches are made in this article. We present some basic algorithms to calculate the main direction of buildings, improve a method based on line approach, introduce a new method based on PCA. Experiments are made to compare them.