Dense monocular Simultaneous Localization and Mapping by direct surfel optimization

This work presents a novel approach for monocular dense Simultaneous Localization and Mapping. The surface to be estimated is represented as a piecewise planar surface, defined as a group of surfels each having as parameters its position and normal. These parameters are then directly estimated from the raw camera pixels measurements, by a Gauss-Newton iterative process. The representation of the surface as a group of surfels has several advantages. It allows the recovery of robust and accurate pixel depths, without the need to use a computationally demanding depth regularization schema. This has the further advantage of avoiding the use of a physically unlikely surface smoothness prior. New surfels can be correctly initialized from the information present in nearby surfels, avoiding also the need to use an expensive initialization routine commonly needed in Gauss-Newton methods. The method was written in the GLSL shading language, allowing the usage of GPU thus achieving real-time. The method was tested against several datasets, showing both its depth and normal estimation correctness, and its scene reconstruction quality. The results presented here showcase the usefulness of the more physically grounded piecewise planar scene depth prior, instead of the more commonly pixel depth independence and smoothness prior.

Feature Pyramid Network Based Efficient Normal Estimation and Filtering for Time-of-Flight Depth Cameras

In this paper, an efficient normal estimation and filtering method for depth images acquired by Time-of-Flight (ToF) cameras is proposed. The method is based on a common feature pyramid networks (FPN) architecture. The normal estimation method is called ToFNest, and the filtering method ToFClean. Both of these low-level 3D point cloud processing methods start from the 2D depth images, projecting the measured data into the 3D space and computing a task-specific loss function. Despite the simplicity, the methods prove to be efficient in terms of robustness and runtime. In order to validate the methods, extensive evaluations on public and custom datasets were performed. Compared with the state-of-the-art methods, the ToFNest and ToFClean algorithms are faster by an order of magnitude without losing precision on public datasets.

Geological Modeling Method Based on the Normal Dynamic Estimation of Sparse Point Clouds

The effect of geological modeling largely depends on the normal estimation results of geological sampling points. However, due to the sparse and uneven characteristics of geological sampling points, the results of normal estimation have great uncertainty. This paper proposes a geological modeling method based on the dynamic normal estimation of sparse point clouds. The improved method consists of three stages: (1) using an improved local plane fitting method to estimate the normals of the point clouds; (2) using an improved minimum spanning tree method to redirect the normals of the point clouds; (3) using an implicit function to construct a geological model. The innovation of this method is an iterative estimation of the point cloud normal. The geological engineer adjusts the normal direction of some point clouds according to the geological law, and then the method uses these correct point cloud normals as a reference to estimate the normals of all point clouds. By continuously repeating the iterative process, the normal estimation result will be more accurate. Experimental results show that compared with the original method, the improved method is more suitable for the normal estimation of sparse point clouds by adjusting normals, according to prior knowledge, dynamically.