Building Contour Detection Based on Straightness Level Set

Abstract

Traditional level set algorithm is not automated enough to extract buildings in UAV remote sensing images. This paper proposes an improved level set algorithm to solve this problem. Firstly, it combined the traditional level set and chain code, then it introduced the parameters of straightness. Finally, it could automatically detect whether the segmentation curve is a building by calculating the straightness. The experimental results show that the straightness of building contour calculation can reach more than 95%, which meets the requirement of straightness judgment, and the value of non-building straightness is relatively low. The algorithm is more automated than the traditional level set, and can automatically eliminate the segmentation curve of the UAV remote sensing image that does not belong to the building.