Wiki title

Photogrammetry

Photogrammetry provides a highly effective technical solution for data acquisition in the context of digital twins by capturing and processing photographic data to create accurate 3D models of physical objects, environments, or systems. This technique is particularly valuable for generating realistic and spatially accurate representations of the physical world, which can be integrated into digital twin platforms for monitoring, analysis, and decision-making.

Key concepts

Photogrammetry involves capturing multiple overlapping photographs of an object or scene from various angles. These images are then processed using specialized software to:

Triangulate Spatial Coordinates: The software determines the 3D position (x, y, z) of points in space by analysing the relationships between overlapping images.
Generate Point Clouds: A dense collection of 3D points is created, representing the surface geometry of the physical object or environment.
Create 3D Models: The point cloud is further processed into textured 3D meshes or high-resolution models that accurately replicate the real-world asset.

In summary, photogrammetry is a versatile and cost-effective solution for data acquisition in digital twins. Its ability to produce detailed and realistic 3D models makes it invaluable across industries such as construction, urban planning, natural resource management, and cultural heritage preservation. When integrated with digital twin platforms, photogrammetric data enhances visualization, monitoring, and decision-making capabilities[1][4][7].

Challenges

Lighting Conditions: Photogrammetry requires good lighting conditions for optimal results; it may struggle in low-light environments compared to LiDAR.
Processing Time: Generating high-quality 3D models can be computationally intensive and time-consuming.
Vegetation Limitations: Unlike LiDAR, photogrammetry cannot penetrate vegetation layers effectively, which may limit its application in dense forested areas[5].

Mechanisms

High-Resolution 3D Modeling

Photogrammetry captures fine details and textures of physical objects or environments, making it ideal for creating photorealistic 3D models.

For example, in construction and infrastructure projects, photogrammetry can model buildings, bridges, or roads with millimetre-level accuracy[1][2][5].

Scalable Data Collection

Photogrammetry can be applied to objects of varying scales, from small components to large environments like cities or industrial plants. This makes it versatile for digital twins across industries such as manufacturing, urban planning, and natural resource management[1][3][7].

Drones equipped with high-resolution cameras are often used to capture data efficiently over large areas[4][7].

Cost-Effective Reality Capture

Compared to other methods like LiDAR, photogrammetry can be more cost-effective as it often uses readily available devices such as consumer-grade cameras or drones[1][4][7].

Open-source or commercial photogrammetry software (e.g., Pix4D, DroneDeploy) processes these images into usable 3D models.

Integration with Digital Twin Platforms

Photogrammetry-generated models can be imported into digital twin platforms (e.g., AWS IoT TwinMaker or Bentley iTwin) to provide a visual context for real-time data streams[4][14].

These models serve as a foundation for integrating sensor data and enabling simulations.

Technical advantages

Accuracy and Realism

Photogrammetry produces highly realistic models by stitching together photographs with detailed textures and colours[5][6]. This visual fidelity is critical for applications like architectural visualization or heritage preservation.

Flexibility

It supports diverse use cases, from modeling industrial equipment in smart factories to mapping entire cities for urban planning[1][3].

Accessibility

Advances in technology have made photogrammetry accessible through consumer-grade devices like smartphones (e.g., iPhone Pro with stereo vision) and drones[1][4].

Georeferencing

By incorporating ground control points (GCPs), photogrammetry ensures that models are geospatially accurate and aligned with real-world coordinates[5].

Examples

Mobile Mapping and AI-Powered Asset Recognition

Skilled Mapping's presentation by Alex Wrigglesworth demonstrates how high-resolution photogrammetric imagery combined with AI can scale data collection and improve digital twin accuracy. Rather than collecting data once, the company uses continuously updated mobile mapping vehicles equipped with LiDAR and high-specification cameras to capture cities comprehensively.

The approach uses high-resolution photogrammetric imagery to extract detailed information about physical assets at scale. When collecting city-wide data (such as Birmingham), the system can automatically identify and classify visible objects using AI. For example, by processing high-resolution street-level imagery, the team can identify every post box in a city and extract its precise location and visual characteristics. As Wrigglesworth explains:

"If someone says we need to know how many post boxes are in Birmingham, we can run the AI and we can say this is how many post boxes there are, this is their location, this is a photo of all those post boxes. But that can actually be used for anything that you can visually see and we can get in high clarity from the image."

The intelligence extraction process works through progressive refinement: operators label visible elements in high-resolution imagery, then train AI models to automatically classify similar features across the entire dataset. Critically, the approach handles the real challenge of continuously maintaining digital twins—rather than assuming one-time data collection suffices, the system enables rapid re-tasking of collection vehicles to update specific locations on demand, ensuring critical changes (such as newly constructed buildings or security-relevant updates) are captured promptly.

Emergency Response and Spatial Analysis

Devon and Cornwall Police's digital twin for the G7 summit demonstrates photogrammetry's role in creating high-fidelity environmental models for complex operational planning. The initiative combined photogrammetry with complementary data acquisition methods to support emergency response coordination.

The police force employed two distinct methods for 3D data capture: photogrammetry and LiDAR. The workflow began with creating a digital surface model from aerial data, which was then draped with high-resolution aerial imagery (procured through government contracts with Ordnance Survey). This photorealistic base layer provided spatial context. Subsequently, drones captured higher-resolution photogrammetric data for areas requiring detailed analysis—enabling precise measurement of features like drains, windows, and building edges where detailed dimensional accuracy was essential:

"We did a digital surface model initially, we draped that with area imagery as part of the government contracts Ordnance survey then came in and built 3D cities for us with a specific quality that was fine for what we needed and then we used drones to do the much kind of higher detail areas where we could start to look at drains, windows, take measurements and do that work."

Unlike purely visual applications, the photogrammetric models were systematically integrated with geospatial intelligence and operational data. The digital twin became actionable through viewshed analysis—determining what security personnel could observe from specific vantage points—combined with linkage to reference materials, emergency protocols, and real-time operational information. The system supported route optimization and risk assessment by enabling first responders to understand ground conditions, elevation changes, and sightlines before arriving on-site.