2D-Image to 3D-Range Scan Registration in Urban Environments

Here we propose a robust way of estimating statistical similarity measurements for 2D-3D data that are collected in various urban scenes with both low-cost and high-end range sensors.

Instead of measuring the similarity of 2D-3D data on specific pair of 2D-3D attributes (e.g. the commonly used intensity-reflectivity), we compute similarity measurements between a set of 2D-3D attribute-pair candidates. Depending on the scenario type (e.g. highway, urban road, etc.), different weighting schemes are applied to combine all similarity measurements into one metric.

The weighting scheme is learned from a small set of example 2D-3D data, which can be registered manually or semi-autonomously with relative small effort. After that, the rest of the 2D-3D data can be registered automatically using our approach. A video illustration can be found here.

Referred publications:

• 2D-Image to 3D-Range Registration in Urban Environments via Scene Categorization and Combination of Similarity Measurements,

Y. Zhao, Y. Wang, Y. Tsai, IEEE Int. Conf. on Robotics and Automation (ICRA), 2016. (paper, presentation, preparation code, registration code)