Plane Detection Based on an Improved RANSAC Algorithm

When obtaining point cloud data of the measured object through 3D scanning, it is inevitable to encounter noise and outliers, which seriously affect the accuracy of estimating point cloud plane parameters and fitting planes. The Random Sample Consensus (RANSAC) algorithm can effectively estimate point cloud plane parameters and fit planes with certain robustness, but it has redundancy as it needs to distinguish inliers from outliers in each iteration, which has a certain impact on running efficiency. This article proposes an improved RANSAC algorithm based on Principal Component Analysis (PCA) method, combined with setting certain criteria to eliminate gross errors and outliers in point cloud data, in order to obtain ideal plane fitting parameters. Experiments show that compared with some traditional algorithms, this method can adapt well to the presence of gross errors and outliers in point cloud data, obtain better estimates of plane parameters, and is a robust plane fitting algorithm.