A Quality-Based Criteria for Efficient View Selection

The generation of complete 3D models of real-world objects is a well-known problem. The accuracy of a reconstruction can be defined as the fidelity to the original model, but in the context of the 3D reconstruction, the ground truth model is usually unavailable. In this paper, we propose to evaluate the quality of the model through local intrinsic metrics, that reflect the quality of the current reconstruction based on geometric measures of the reconstructed model. We then show how those metrics can be embedded in a Next Best View (NBV) framework as additional criteria for selecting optimal views that improve the quality of the reconstruction. Tests performed on simulated data and synthetic images show that using quality metrics helps the NBV algorithm to focus the view selection on the poor-quality parts of the reconstructed model, thus improving the overall quality.