Article directory
1. Introduction
In CGAL, this method takes a set of point sets that cannot be vectored as input and outputs a set of point clouds of specified shape. where the shape is specified by the ransac type. The basic RANSAC method repeats the following steps:
- Randomly select samples from the input points;
- fit a shape to the selected sample;
- Calculates the number of interior points of a shape to determine whether the interior point cloud is within the user-specified shape error tolerance.
Steps 1-3 are repeated for a specified number of iterations, and the shape with the largest number of internal points is retained, called the largest shape.
In our description, the error between a point and a shape is defined by its distance and normal deviation from the shape. The random subset is chosen based on the minimum number of points (with normals) required for the primitive we define.
For very large point sets, the original RANSAC method is impractical when testing all possible candidate shapes against the input data to find the largest shape. So an efficient RANSAC method is implemented in CGAL, the main idea behind this method is to test candidate shapes against a subset of the input data. Shape candidates are constructed until the probability of missing the largest candidate is below a user-specified threshold. Extracting the largest shape is repeated until no more shapes, constrained to cover the minimum number of points, will be extracted.
The support of a shape is the footprint of the points covered by the primitive. To avoid generating