A Study of Feature Points Extraction Based on Point Cloud Data Sets and Model Simplification in Goaf

The shape measurement of goaf is an important job in mine project. With the rapid development of the measuring equipment, 3D laser scanner, which makes the efficiency of traditional work being improved, is widely used in goaf. The final result of 3D laser scan is a huge collection of points. Although different measuring intervals will lead different point cloud density, it also needs a large storage of several MB at least for a single gob area. Point clouds can be used to describe a spatial object’s all details with great precision, but there is still a large amount of redundant data. In mathematics, the curvature is the reciprocal of the radius, and a smaller curvature implies a large radius. Intuitively, curvature is the amount by which a geometric object deviates from being flat, or straight in the case of a line; it is defined in different ways depending on the context. Based on the point cloud datasets of a goaf, this thesis offers a new method for modelling the gob area by using the characteristic of curvature of planar curve and the local curvature of spatial curved surface. Actually, the feature points and feature regions of a goaf will be expressed via the curvature parameter. Aiming at simplifying the model, a mathematical model is constructed in order to identify the feature region and non-feature region automatically. The curvature parameter, which is an important controllable variable, can be used to help judging the feature region and non-feature region. Actually, the above-mentioned method could even construct the 3D model of gob area automatically via sampling more points as the feature points in feature region and fewer points in non-feature region. At last, the thesis compares the models that are simplified via different parameters and analyzes how the simplified parameters influence models from the point of system storage space and the extent of distortion etc.