3D modelling method and application to a digital campus by fusing point cloud data and image data

Area of Science:

• Geomatics Engineering
• Computer Vision
• 3D Modelling

Background:

• Single-source 3D modelling faces challenges like deformation and fuzzy textures due to insufficient imagery.
• Airborne laser scanning (LiDAR) generates massive datasets, demanding significant computational resources for processing.
• Existing methods struggle with high-quality, high-precision city-level 3D model construction.

Purpose of the Study:

• To develop a novel 3D modelling method by fusing data from inclined photogrammetry and LiDAR.
• To address the limitations of individual data sources in creating detailed and accurate 3D building models.
• To propose a technical process for multi-source data fusion in 3D reconstruction.

Main Methods:

• Utilized a university library building for case study.
• Implemented multi-source data fusion through spatial alignment and coordinate system unification.
• Applied the iterative closest point (ICP) algorithm for precise point cloud alignment and integration.

Main Results:

• Achieved significant improvements in 3D model accuracy and detail.
• Effectively repaired model deficiencies and optimized fine details.
• Enhanced the quality of photogrammetry models, preserving geometric and texture information.

Conclusions:

• The proposed fusion method offers superior 3D reconstruction capabilities.
• Demonstrated high positional accuracy for integrated inclined photogrammetry and LiDAR data.
• Provides innovative approaches for digital campus construction and other engineering applications.