Surface Reconstruction Based on Neural Networks

Surface reconstruction is an important trend in 3D scanning. The problem is to recreate surfaces from a given point cloud within the shortest possible time and with a given quality criteria. There is a set of different approaches for solving this problem, which includes Self-Organized Maps, Bayesian reconstruction and Poisson reconstruction.

The aim of the research is to find the most suitable method based on Machine Learning unsupervised learning techniques for reconstruction of interior and exterior 3D scans of original objects. The self-organizing map type of the artificial neural network was chosen due to its ability to produce good results without restrictions on point cloud size and points’ order. It can generate meshes from a small number of point samples in an unorganized point cloud.

The purpose of this paper is to analyze and compare results obtained with the usage of two self-organizing map types – Surface Growing Neural Gas (sGNG) and Growing Cell Structures (GCS) reconstruction – for reconstruction of a 3D mesh from point cloud. In addition, the sGNG extension to multithreading is presented and compared to the original approach.

Read the full article here:

AMC Bridge Webinars

Industry experts share their knowledge, advice, and insights

AMC Bridge news

  • Media about us

    May 13, 2026

    AMC Bridge Demos AI‑Driven Similar Parts Search

    AMC Bridge’s latest technology demonstration, Similar Parts Search, explores how combining artificial intelligence (AI) with advanced 3D geometric processing can address these challenges and improve the discovery, reuse, and evaluation of structurally similar parts across engineering workflows.
  • Technology demonstration

    May 04, 2026

    AMC Bridge Demonstrates AI‑Driven Similar Parts Search Using Graph Neural Networks and 3D Geometry

    Engineering and manufacturing organizations continue to face challenges when managing large repositories of 3D models, including manual shape comparison, inconsistent similarity assessments, and the time‑consuming navigation of extensive part libraries. Traditional CAD search and manual similarity methods rely heavily on metadata, naming conventions, or subjective human judgment—approaches that do not scale and often produce inconsistent results. AMC Bridge’s latest technology demonstration, Similar Parts Search, explores how combining artificial intelligence (AI) with advanced 3D geometric processing can address these challenges and significantly improve the discovery, reuse, and evaluation of structurally similar parts across engineering workflows. By leveraging learned geometric representations of 3D models, the POC enables consistent, geometry‑aware similarity detection across large repositories, independent of how parts are named or categorized.
  • Partnerships

    Apr 29, 2026

    What It Takes to Build Connected Engineering Systems at Scale

    When the founders of AMC Bridge formed our company a quarter century ago, we wanted to help clients with digital transformation – but we realized that technical know-how alone wasn’t enough. We had to develop a deep understanding of our customers – their ambitions, their competition, their existential challenges. What we’ve learned from hundreds of clients while engineering their software ecosystems holds true across enterprises. To move decisively, modern organizations need a secure, open platform that connects data, automates high-value work, and gives stakeholders clear, current context for their next moves.
Discover