A method of correction building roofs offset using wall baselines from SAR imagery

In the imaging process of optical image, there is an offset of building roof footprint in the image from the location of the real world because of the angle of sensor. The offset as one of the important elements establishment of urban basic geographic information has an impact on city planning. In this paper, we proposed to correct the building roof footprint offset in optical image using wall baselines extracted by morphological attribute profiles(MAPs) algorithm from SAR imagery. Firstly, MAPs algorithm was adopted to detect building connected components(CCs) from high-resolution SAR imagery, considering the ability of MAPs on detecting buildings of different sizes, shapes and directions. Then, the building wall baselines were extracted according to the skeleton thinning lines by applying alternating sequence morphological thinning on buildings CCs. Next, we built a line feature matching template based on similarity and distance for template matching calculation of building roof footprint characteristics extracted from optical image. Finally, building roof footprints offset was corrected pixel by pixel. The results show that the proposed method is feasible and the correction rate of building roof offset is improved compared with the shadow-based method