Source-linked AI summary
3D Point Cloud from Close-Range Photogrammetry for Defect Characterisation of Rubberised Concrete
Jiacheng Liu, Mohammed Alnahhal, Ailar Hajimohammadi, Sara Gonizzi Barsanti, Jinling Wang, Mohsen Kalantari
TL;DR
Fine-scale crack analysis in rubberised concrete is difficult because conventional LiDAR systems cannot reliably resolve microcracks, while photogrammetric workflows remain insufficiently validated for this heterogeneous material. The study develops and evaluates a close-range SfM/MVS workflow using high-resolution imagery, RGB-guided crack extraction, and pre-/post-test deformation comparison. The DSLR reconstruction achieved sub-millimetre effective resolution, and the workflow represented crack morphology and surface displacement in laboratory specimens.
Problem
Conventional LiDAR is limited for fine-scale defect analysis because its laser spot size can exceed microcrack width, while photogrammetric validation for rubberised concrete remains limited.
Method
The study develops a systematic close-range photogrammetry workflow using SfM/MVS reconstruction, calibrated high-resolution imagery, RGB-guided crack extraction, and pre-/post-test deformation analysis.
Results
The Canon DSLR reconstruction achieved 0.05 mm average point spacing with 0.15 mm GSD-limited effective optical resolution, while measured crack width was approximately 2.3 mm versus 2.36 mm physically.
Takeaways & Limitations
Close-range photogrammetry provides a flexible, high-resolution alternative to LiDAR for laboratory surface inspection, defect morphology representation, and deformation monitoring.
Takeaways & Limitations
TLS laser spot sizes typically range from 3 mm to 5 mm, limiting minimum detectable crack width and bridging sub-millimetre microcracks.
Abstract
from arXiv · showhide
While three-dimensional (3D) point clouds are widely used in civil engineering, mainstream LiDAR systems such as Terrestrial Laser Scanning (TLS) are physically constrained to laboratory environments. Since their laser spot size typically exceeds the width of microcracks, the beam physically bridges over voids, rendering TLS unsuitable for fine-scale defect analysis. Alternatively, close-range photogrammetry utilising Structure-from-Motion (SfM) and Multi-View Stereo (MVS) algorithms offers a solution for testing highly tortuous materials, and its utility at fine-scale remains underexplored. This study adapts photogrammetric workflows specifically for rubberised concrete (RuC), a sustainable composite exhibiting high ductility and complex fracture morphologies. High-resolution image sets were captured using a Canon DSLR and an iPhone 16 to generate dense 3D models. Comparisons revealed that the DSLR-based reconstruction achieved sub-millimetre resolution, demonstrating superior performance for fine-scale surface monitoring. An RGB-guided crack extraction method was developed to enhance the identification of surface defects and isolate potential crack areas from the background. The extracted crack regions were visually distinguishable and provided a well-structured geometrical representation of defect morphology. Furthermore, a Pre and Post-Test deformation analysis was conducted to quantify surface displacement across testing stages. The results confirm that this close-range photogrammetry workflow is a flexible, high-resolution alternative to LiDAR for surface inspection and deformation monitoring of specimens in laboratory settings. Ultimately, this approach establishes a robust geometric baseline for future automated 3D feature characterisation and material performance evaluation.
1. Introduction
Rubberised concrete offers sustainability and improved ductility but has complex, fine-scale fracture behaviour that is difficult to characterise with conventional inspection and point-cloud methods. This study validates a systematic close-range photogrammetry workflow for sub-millimetre defect reconstruction and quantitative analysis.
- Motivation: Rubberised concrete uses recycled tyre rubber, improving ductility, impact resistance, and energy absorption while often reducing strength capacity.Its optimal balance of rubber content, particle size, and admixtures remains under investigation.
- Research problem: Cracks are key indicators of rubberised concrete performance under flexural loading, but millimetre-scale defects are difficult to quantify completely in three dimensions.Visual inspection and contact measurements are labour-intensive and do not capture complete fracture-surface topology.
- Research problem: Existing computer-vision, LiDAR, and Structure-from-Motion methods support civil-engineering inspection but remain limited for fine-scale fracture surfaces and incipient cracks.Macro-scale point-cloud methods provide flexibility for structural classification and health monitoring, whereas fine defects require millimetre-level accuracy.
- Study objectives: A systematic high-resolution 3D reconstruction workflow is proposed for laboratory specimens, targeting metric accuracy at the sub-millimetre level.The workflow addresses data acquisition and processing rather than relying only on post-test visual inspection.
- Study objectives: The generated point clouds support crack-length estimation, deviation analysis, enhanced visual communication, and future publicly available research use.These applications extend the workflow from reconstruction toward quantitative defect analysis.
2. Related work
Civil-engineering point clouds increasingly support structural inspection, but conventional LiDAR is constrained for laboratory-scale microcrack analysis. Close-range photogrammetry offers a promising alternative, although systematic validation for heterogeneous rubberised concrete remains limited.
- Existing point-cloud applications: TLS and MLS point clouds support infrastructure monitoring, deformation assessment, defect extraction, quantification, and building information modelling.Their main advantage is efficient capture of comprehensive three-dimensional spatial geometry.
- Existing point-cloud applications: Point-cloud quality-control systems have demonstrated automated assessment of geometric irregularities, positional accuracy, surface flatness, and crack presence in precast concrete.Related approaches use K-nearest-neighbour classification and Delaunay triangulation for geometry-driven defect measurement.
- Existing point-cloud applications: Deep-learning and computer-vision methods have been applied to point-cloud inspection, including rebar segmentation and benchmark datasets for concrete surface-defect detection.One reported rebar framework achieved over 90% accuracy and more than 97% recall for main and tie rebars.
- LiDAR limitations: LiDAR systems are effective for macro-scale structural monitoring but face a physical bottleneck when applied to fine-scale laboratory defects.TLS laser spot sizes typically range from 3 mm to 5 mm, causing beams to bridge sub-millimetre microcracks.
- Close-range photogrammetry: Close-range photogrammetry uses SfM and MVS to generate dense, high-fidelity point clouds from overlapping 2D images with sub-millimetre-level detail.Its image-based reconstruction also provides more accurate RGB information for capturing subtle geometric and textural variations.
- Research gap: Systematic validation of photogrammetric workflows in laboratory environments remains limited, particularly for linking 3D models with rubberised-concrete material behaviour and mix designs.Rubberised concrete’s heterogeneous composition and distinctive cracking behaviour provide a challenging validation setting.
- Research gap: The study addresses this gap by investigating the performance, applicability, and accuracy of close-range photogrammetric reconstruction for surface-defect analysis in rubberised-concrete specimens.The stated focus is a validated, systematic, and efficient workflow for characterising minor surface defects.
3. Methodology
The methodology establishes a controlled close-range photogrammetry workflow for reconstructing small rubberised-concrete specimens and producing metrically reliable 3D data. It combines calibrated image acquisition, scale references, SfM processing, and dense point-cloud generation.
- 3. Methodology: RuC prism beams measuring 550 mm × 100 mm × 100 mm were tested to failure under four-point flexural bending.The mix included river sand, 10 mm basalt sand, and 5 mm shredded rubber from waste tyres.
- 3. Methodology: A systematic acquisition protocol fixed camera stability, imaging geometry, lighting, and scale references to support consistent geometric and optical reconstruction.The protocol used a tripod, calibration targets, fixed object-to-camera distances, and artificial lighting.
- 3. Methodology: Images were captured with a Canon EOS 5D Mark IV DSLR and an iPhone 16 to compare reconstruction quality across image sources.The DSLR used a 24 mm prime lens, while the iPhone used a 26 mm equivalent focal length.
- 3. Methodology: 61 images with 6720 × 4480-pixel resolution and at least 60% overlap were collected to provide comprehensive surface coverage and robust tie-point matching.The images were captured from multiple viewing angles around the specimen.
- 3. Methodology: Metashape reconstruction comprised camera calibration, image alignment, scaling and optimisation, and dense point-cloud generation.Checkerboard calibration corrected lens distortion, while control scale bars converted relative geometry into physical dimensions and bundle adjustment refined camera geometry.
4. Experiment results and discussions
The photogrammetric workflow produced geometrically reliable, sub-millimetre point clouds, supported RGB-guided crack extraction, and quantified millimetre-scale deformation between testing stages.
- 3D model quality evaluation: 2.3 mm reconstructed crack width closely matched the 2.36 mm physical measurement.The comparison supported the geometric precision and scaling reliability of the reconstructed model.
- 3D model quality evaluation: 0.05 mm average point spacing for the Canon DSLR compared with 0.23 mm for the iPhone.The effective optical resolution remained bounded by a 0.15 mm ground sampling distance.
- Crack identification: RGB-guided grayscale and logarithmic contrast enhancement isolated darker crack areas from intact surface points.The processed radiometric contrast enabled spatial isolation of fracture points for subsequent geometric analysis.
- Crack identification: The extracted crack points retained a well-structured geometric representation suitable for precise morphological interpretation.A skeleton-based algorithm was identified for subsequent simplification into a measurable spatial network.
- Pre- and Post-Test deformation analysis: 2.2 mm average surface displacement occurred in the primary fracture zone during the four-point bending test.Most deviation values ranged from 1.3 mm to 2.9 mm, with the maximum displacement corresponding to visible cracking.
5. Conclusion and future work
The study demonstrates close-range photogrammetry as a sub-millimetre alternative for analysing minor defects and deformation in rubberised concrete. It concludes that the workflow supports crack segmentation, dimensional estimation, and millimetre-scale displacement measurement, while identifying automation and advanced reconstruction as future directions.
- Conclusion: Close-range photogrammetry generated dense point clouds with sub-millimetre resolution for minor-defect and deformation analysis in rubberised concrete.
- Conclusion: The workflow compared reconstructions with physical measurements, segmented cracked regions, and quantified millimetre-scale surface displacements.
- Future work: Future work will compare NeRF and 3D Gaussian Splatting with SfM and integrate handheld LiDAR for metric-scale validation.
- Future work: Future development will automate 3D defect feature extraction and monitor dynamic crack propagation using multi-camera or videogrammetry techniques.