Low-rank Based Algorithms for Rectification, Repetition Detection and De-noising in Urban Images

In this thesis, we aim to solve the problem of automatic image rectification and repeated patterns detection on 2D urban images, using novel low-rank based techniques. Repeated patterns (such as windows, tiles, balconies and doors) are prominent and significant features in urban scenes. Detection of the periodic structures is useful in many applications such as photorealistic 3D reconstruction, 2D-to-3D alignment, facade parsing, city modeling, classification, navigation, visualization in 3D map environments, shape completion, cinematography and 3D games. However both of the image rectification and repeated patterns detection problems are challenging due to scene occlusions, varying illumination, pose variation and sensor noise. Therefore, detection of these repeated patterns becomes very important for city scene analysis. Given a 2D image of urban scene, we automatically rectify a facade image and extract facade textures first. Based on the rectified facade texture, we exploit novel algorithms that extract repeated patterns by using Kronecker product based modeling that is based on a solid theoretical foundation. We have tested our algorithms in a large set of images, which includes building facades from Paris, Hong Kong and New York