TEXTURE MAPPING FOR BUILDING FACADES WITH REPEATED PATTERNS USING TERRESTRIAL IMAGERY AND LIDAR DATA

摘要

Texture mapping for building facades is an important work in the application of photorealistic building models. The terrestrial imagery provides a wealth of facade information with high spatial resolution for texture mapping. However, the facade may be occluded by other objects, such as trees, cars, and so on, in image acquisition. Thus, this paper integrates terrestrial images and LIDAR data to remove the occlusions in texture mapping. Considering that many building facades are with repeated patterns, we select the optimal facade texture for the similar repeated patterns. The major works contain data registration, occlusion detection and compensation, and facade texture selection. After the data registration for building models, images, and LIDAR data, the proposed method detects occlusion areas in images by the comparison of building models and LIDAR data. The occlusion parts are to be compensated by other different view-angle images. The grey level co-occurrence matrix (GLCM) analysis is employed for the selection of the most appropriate facade texture. Experimental results show that the proposed method may select the optimal texture for the repeated patterns facades. It provides high fidelity textures in the generation of photorealistic building models.