Reconstruction of complete 3D models from multi-view point clouds

Finished: 2019-05-23

BSc assignment

Using a stereo camera, a dense 3D point cloud of an object can be reconstructed. However, looking at an object from only one angle is does not give nearly enough information to create a complete 3D model.Therefore, will have to be looked at from multiple angles. This will give multiple point clouds that need to be merged, which does result in a complete 3D model.

In this assignment, a mathematical workflow and algorithm will be developed in order to have a complete 3D model. The main features of this algorithm should be noise filtering, merging multiple point cloud, remove imperfections and creating a mesh surface, which will also be the order in which the algorithm is performed. Using this algorithm a complete 3D model can be made by just imputing the data recorded by the stereo camera into the algorithm.

The application for this is to observe 3D changes after burn of the skin. For example a doctor could than reconstruct an 3D model on his computer and inspect changes to the skin. The algorithm will be tested by doing experiments on the human body, such as arms and legs.