The Influence of the Input Parameters Selection on the RANSAC Results

Abstract

The RANSAC (RANdom SAmpling and Consensus) enables us to search within a given group of points for subgroups of points that belong to a mathematically describable object or a part of an object. The number of iterations within a single repetition depends on the data, selection and settings of the input parameters (percentage of inliers, probability and minimum number of points that uniquely define a geometrical shape). In our research we applied simulation modelling to analyse the influence of the selection of input parameters on the approximation of the sphere and plane models.