Local-to-Global Point Cloud Registration using a Viewpoint Dictionary

Local-to-Global Point Cloud Registration using a Viewpoint Dictionary
December, 12, 2017
in Room 1061 Electrical Eng. Building Technion City

Pixel Club

Speaker: David Avidar

Affiliation: Dept. of Electrical Engineering Technion



Local-to-Global Point Cloud Registration using a Viewpoint Dictionary

Local-to global point cloud registration is a challenging task due to the substantial differences between these two types of data, and the different techniques used to acquire them. Global clouds cover large-scale environments and are usually acquired aerially, e.g., 3D modeling of a city using Airborne Laser Scanning (ALS). In contrast, local clouds are often acquired from ground level at a much smaller range, using Terrestrial Laser Scanning (TLS). The differences are typically manifested in their point density distribution, occlusions’ nature, and measurement noise characteristics. As a result of these differences, existing point cloud registration approaches, such as keypoint-based registration, tend to fail.
We propose a novel registration method that is robust to the different characteristics of global and local point clouds. The method is based on converting the global cloud into a viewpoint-based dictionary. We associate each viewpoint with a panoramic range-image, capturing the geometry of the visible environment. Then, plausible local-to-global transformations can be found via a dictionary search, i.e., finding the best matches between the local and dictionary panoramic range-images. We show efficient dictionary search can be done in the Discrete Fourier Transform (DFT) domain, using phase correlation. An efficient registration refinement method for urban environments, based on converting the point clouds into edge-maps, is also presented.
We demonstrate that the proposed viewpoint-dictionary-based registration method is superior to a state-of-the-art, keypoint-based method (FPFH – Fast Point Feature Histogram), even without any GPS measurements. For the evaluation, we used a challenging dataset of 108 TLS local clouds and an ALS large-scale global cloud, in a 1〖km〗^2 urban environment.

*    סטודנט לתואר שני בהנחיית פרופסור דוד מלאך ודר’ מאיר בר-זוהר.