For the weekly seminar of the STORM research group, Nicolas Mellado will present his current research works on Transformation Space Sampling and Analysis for Point-Cloud Registration.
Data acquisition in large-scale scenes regularly involves accumulating information across multiple scans. A common approach is to locally align scan pairs using Iterative Closest Point (ICP) algorithm (or its variants), but requires static scenes and small motion between scan pairs. This prevents accumulating data across multiple scan sessions and/or different acquisition modalities (e.g., stereo, depth scans). Alternatively, one can use a global registration algorithm allowing scans to be in arbitrary initial poses.
In this talk, I will review latest advances on point-cloud global registration, focusing on so-called sampling approaches, where the space of rigid transformations is sampled to find the best alignment between two shapes. I will also present actual challenges related to this topic, and discuss on-going work and promising research directions.