Feature surface extraction and reconstruction from industrial components using multistep segmentation and optimization

The structure of industrial components is diversified, and extensive efforts have been exerted to improve automation, accuracy, and completeness of feature surfaces extracted from such components. This paper presents a novel method called multistep segmentation and optimization for extracting feature surfaces from industrial components. The method analyzes the normal vector distribution matrix to segment feature points from a 3D point cloud. The point cloud is then divided into different patches by applying the region growing method on the basis of the distance constraint and according to the initial results. Subsequently, each patch is fitted with an implicit expression equation, and the proposed method is combined with the random sample consensus (RANSAC) algorithm and parameter fitting to extract and optimize the feature surface. The proposed method is experimentally validated on three industrial components. The threshold setting in the algorithm is discussed in terms of algorithm principles and model features. Comparisons with state-of-the-art methods indicate that the proposed method for feature surface extraction is feasible and capable of achieving favorable performance and facilitating automation of industrial components.

The structure of industrial components is diversified, and extensive efforts have been exerted to improve automation, accuracy, and completeness of feature surfaces extracted from such components. This paper presents a novel method called multistep segmentation and optimization for extracting feature surfaces from industrial components. The method analyzes the normal vector distribution matrix to segment feature points from a 3D point cloud. The point cloud is then divided into different patches by applying the region growing method on the basis of the distance constraint and according to the initial results. Subsequently, each patch is fitted with an implicit expression equation, and the proposed method is combined with the random sample consensus (RANSAC) algorithm and parameter fitting to extract and optimize the feature surface. The proposed method is experimentally validated on three industrial components. The threshold setting in the algorithm is discussed in terms of algorithm principles and model features. Comparisons with state-of-the-art methods indicate that the proposed method for feature surface extraction is feasible and capable of achieving favorable performance and facilitating automation of industrial components.