Have you ever scanned a large-size sculpture, or object, and the measurement results are a total disaster? Scanning large-size objects can be challenging as errors accumulate over a certain distance during measurement.
How can we eliminate accumulated errors? By integrating photogrammetry with 3D scanning, we can enhance the speed and accuracy of data collection and the 3D modelling process.
The word “Photogrammetry” was coined by Prussian architect Albrecht Meydenbauer in his article “Die Photometrographie” in 1867. Photogrammetry develops from plane table, analogy, analytical to digital photogrammetry. Each phase of its development extends about 15 years. In this blog, we mainly focus on digital photogrammetry.
Photogrammetry is a measuring technique that takes photographs from different perspectives to obtain 3D coordinates. Specifically speaking, it extracts geometric information by triangulating the locations of points on the subject.
Photogrammetry software can find characteristic points that are repeated in photos. The distance of these points can be inferred using triangulations. The more photos we take from different angles, the more accurate these locations are. Finally, these points would be converted into a mesh and a 3D model.
Photogrammetry can utilize images from aerial photography among other sources, offering enhanced accuracy and depth by analyzing multiple photographs taken from various vantage points, not limited to aerial views. While aerial photography serves to capture images providing a general overview of terrain, photogrammetry delves deeper, offering meticulous measurements of distance, area, and direction, surpassing the inherent positional errors found in traditional aerial imagery.
The essence of photogrammetry lies in its utilization of multiple overlapping photographs taken from diverse vantage points and angles. Through intricate analysis and correlation of these images, photogrammetry reconstructs a 3D representation of the scene, extracting precise spatial data with remarkable fidelity.
This method finds extensive application across various disciplines such as architecture, engineering, surveying, and quality control, where exactitude in measurement is paramount. By leveraging the synergy between advanced imaging techniques and computational algorithms, photogrammetry emerges as an indispensable tool for capturing and quantifying the intricacies of our physical environment with unparalleled accuracy and efficiency.
In metric photogrammetry, engineers prioritize precision, leveraging necessary points over pixel counts for tasks like quality control and reverse engineering. This method extracts accurate measurements and positions from images using precision algorithms and key points to build models.
Photogrammetry delivers exact, efficient solutions for surveying, designing, analyzing, and documenting. It’s vital in engineering for creating detailed maps, models, and orthophotos for infrastructure and environmental assessments. It also produces accurate documentation of existing structures, aiding renovations and maintenance.
Additionally, photogrammetry enhances design and visualization with 3D models, identifies conflicts, and cuts costs. It monitors construction, checking for deviations to maintain quality and safety.
The 3D laser scanner uses lasers to scan the geometry of an object and obtain its 3D data. Once all the points are captured, a dense point cloud is generated, which can be used to create a 3D model.
Most 3D scanners on the market are capable of scanning objects whose sizes are within a range of 1 m, while it is difficult for these 3D scanners to scan large-scale objects like wind turbines, planes, and buildings. There is where photogrammetry comes in.
When measuring an object with photogrammetry, the first step is to put reflective markers on the object surface and coded targets around it. A scale bar is also necessary to serve as a reference. Then, shots are taken from different perspectives, while making sure that you take photos that overlap.
These photos will help to construct a general 3D geometry of the object. The details of the object’s surface then can be captured using a 3D scanner.
With high-resolution and full-frame cameras, a photogrammetry system can give you the highest quality output. Thanks to its large shooting area and precise algorithms, it can reduce errors of connection that accumulate over distance.
Scantech offers MSCAN photogrammetry system for scanning large-scale objects, with high requirements in accuracy and measurement repeatability. MSCAN photogrammetry system can work alone, or work with handheld 3D scanners, to achieve a volumetric accuracy of up to 0.015 mm/m.
MSCAN
KSCAN
To take photogrammetry to a new level, Scantech provides KSCAN composite 3D scanner. It brings infrared laser scanning, blue laser scanning, and photogrammetry into one single device.
It delivers highly repeatable scanning results with great details in high efficiency. When it comes to measurement, it is suitable for both large-scale and tiny objects.
Here is an example of how KSCAN 3D scanner helps to inspect a planet carrier gear for a wind turbine. The diameter of this kind of planet carrier gear is generally bigger than 1 meter and weighs more than 1 metric ton.
Usually, planet carrier gears are produced in small batches as the types of planet carrier gears vary. Engineers can get every single detail of the part and obtain precise 3D measurement results, using KSCAN-Magic 3D scanner with built-in photogrammetry,
To optimize photogrammetry outcomes, a carefully curated set of photos is crucial, providing enough information to deduce the required data. While a camera captures a scene in 2D, photogrammetry analyzes these images to create a 3D model. Sufficient overlapping images are needed to reconstruct a comprehensive 3D model. However, the number of photos required depends on the object’s complexity and project needs. Key photogrammetry algorithms include:
Feature Matching Algorithms identify and aligning corresponding features across overlapping images that meticulously detect distinctive features like corners, edges, or keypoints within the images, facilitating the establishment of correspondences between points in different images. By accurately matching features, photogrammetric systems can subsequently triangulate and reconstruct the three-dimensional structure of the scene with precision and fidelity.
Bundle Adjustment refine the estimated positions and orientations of cameras in a photogrammetric setup. By minimizing disparities between observed and predicted image features, bundle adjustment ensures the coherence and accuracy of the reconstructed 3D scene.
Triangulation algorithms form the cornerstone of 3D reconstruction in photogrammetry. These algorithms compute the three-dimensional coordinates of points in the scene by intersecting rays projected from corresponding points in multiple images. Through triangulation, photogrammetric systems can accurately reconstruct the spatial geometry of objects and surfaces, enabling the creation of detailed and realistic 3D models.
DEM generation algorithms estimate the elevation of terrain points by analyzing stereo image pairs. Methods such as stereo matching, dense image matching, or structured light scanning are used to reconstruct the topography of the terrain.
Feature Extraction algorithms identify and extract specific objects or structures of interest from photogrammetric data. These algorithms can detect and delineate buildings, roads, vegetation, or other cultural and natural features, depending on the application requirements. Feature extraction facilitates the extraction of valuable information for tasks such as urban planning, environmental monitoring, and disaster response.
While the fundamental principles of photogrammetry remain consistent, there are two principal categories pertaining to its application which largely depend on the individual requirements of a project.
The realms of cultural heritage preservation, architectural conceptualization, product design schematics, and virtual reality simulations necessitate the use of photorealistic rendering tools to realistically depict objects from real-world scenarios. As a rule of thumb, highly-detailed models incorporating a multitude of pixels amplify the quality and precision.
Incorporating color for 3D modeling via Photogrammetry consists of procuring an array of high-fidelity colored images capturing diverse perspectives to collectively recreate a comprehensive impression or depiction. Following this step is the utilization of advanced photogrammetric software systems that amalgamate these captured visuals into meticulously detailed 3D structures or models. Subsequently integrating color details culled from initial captures imparts texture onto these computed models aiding in authentically reiterating the primary subject’s appearance.
The final shape of the model is often imperfect. General 3D scanners can sometimes struggle with shiny, transparent or black surfaces, but with photogrammetry the number of artifacts and noise you inevitably have to deal with is much greater. The end result is a model with HD textures, but also with a lot of noise and imperfect geometry.
After the geometric reconstruction is complete, the color information from the original image is applied to the model’s surface texture. This involves mapping colors in an image to corresponding points in a 3D model, thereby creating a visually realistic representation of an object or scene.
