Building Change Detection using Object-Oriented LBP Feature Map in Very High Spatial Resolution Imagery

Building change detection has always been a popular direction in field of remote sensing. The building detection method for very high spatial resolution (VHR) imagery proposed in this paper is based on the classical local binary algorithm. First, histogram equalization and bilateral filtering are used on high-resolution remote sensing images to enhance the contrast and the building edge, which is beneficial to the next process. In the first process, the low-density feature map is obtained through the classical local binary patterns (LBP) algorithm, and then the ground objects are divided into objects by mean shift based on the feature map. This method can accurately segment the boundary of buildings. In the second process, a new rotation uniform invariant local binary pattern algorithm is applied to obtain OOLBP features. Finally, support vector machine classifier (SVM) is adopted for classification. Last, the change types of buildings were identified, including the newly added buildings, building disappearance and building reconstruction. the results show that the overall accuracy and recall ratio exceeds 94%.