More and more drone pilots are getting into the applications of photogrammetry (mapping and 3D modeling). We thought we should help the beginners to start this journey so we created this series. Those articles whose titles start with “Quick Start” are meant for absolute beginners. We try to explain this profession as simply as possible so you could create nice maps and 3D models as a hobby.
Keep in mind that these articles are made for beginners. So, if you’d like to use these articles for commercial and business purposes, we advise you to really dig deeper into this profession.
Through every “Quick Start” article we use free software only, so everyone can try all the presented methods by themselves. 🙂
WHAT IS PHOTOGRAMMETRY?
As I’ve said, I’ll keep it simple and show only the operation of the most common SfM (Structure from Motion) + SIFT (Scale Invariant Feature Transform) algorithms used for 3D modeling… Photogrammetry is a type of remote sensing. The point is to take photos with a camera and then determine the same points on the overlapping images during a separate processing. Determining the points shows where the camera is located relative to the subject being photographed. This process can be done using photos taken with a cell phone camera. A modern processing software even takes the drone’s actual coordinates (location data) into account. The processing software performs various calculations that result in point cloud, textured 3D model, orthophoto map, digital terrain model file. You may notice that this many, many calculations can be very resource intensive, so you need to run the evaluation of the images on a PC with adequate hardware capacity. With the help of photogrammetry, we can examine the geometric, radiometric (measurement of the radiation of electromagnetic waves) and semantic (content) properties of the objects to be imaged. The great advantage of photogrammetry is that a large data set can be obtained quickly, with a dense sampling procedure from a specific area.
THE LIMITS OF PHOTOGRAMMETRY
Photogrammetry as a data collection method also has its limitations, which you should always keep in mind. Through the process we reproduce geometries based on photos. The processing software determines the associated pixels. So, the primary goal is to get the photos in good quality, that is, sharp and well-exposed. The processing software won’t be able to interpret a blurred, noisy image. Transparent, homogenous, glossy and reflective surfaces are difficult, or in some cases impossible to reconstruct. What could they be? For example glass, mirror, water, fresh snow, a well-polished metallic surface, a car and more.
However, if you are smart, there are solutions. You can matte a reflective, glossy surface, for which a so-called scanning spray already exists. Several of these are available. You should take care of the spray to give optimal contrast and not to have too thick a layer (below ~ 10 µm is appropriate). One more thing… If it’s not the goal to leave the subject matte forever, you may want to buy a vanishing spray.
You can use different felt-tip pens, paints and different illumination on homogeneous surfaces. The point is to break the monotonous pixel set that the processing algorithm cannot interpret. The different colored lines and their intersections help the software work to find related pixels more easily.
Coating the surfaces means that you will only have geometric information about the object, as the information content of the original color will be lost.
What is the essence of photogrammetry? What is that one thing that the whole process can’t work without? The light. In the absence of light, photogrammetric measurement is not possible. Visual conditions can affect image capture and quality in the same way, as rain, fog, or dense vegetation can block the camera’s field of view or limit the light needed for clear photography.
So, you should always provide a good quality light source, but the photographed surface should not shine or reflect as it can cause a problem during processing!
THE CATEGORIZATION OF PHOTOGRAMMETRY
In this chapter, I present the categorization of photogrammetry to get you into the picture with the basic concepts. According to the criteria, it will be perceptible that this is a very large and extensive field of science.
The history of photogrammetry dates back almost to the second half of the 1800s. It’s almost as old as photography. In the early stages, ground recordings were used, mainly for architectural purposes (architectural photogrammetry). Images can be grouped according to how they are created and processed:
- Analog – Photographs were processed with analog instruments until the 1970s. Only optical-mechanical instruments existed, where photographs (negatives) were evaluated using special projectors.
- Analytical – The analog method was replaced by computer-controlled analytical instruments, but the photographs were still analog photographs. A significant change compared to analog photogrammetric instruments is that the result of the evaluation here is a digital file that can be processed with CAD (Computer-aided design) software.
- Digital – Today, both the images to be processed and the processing tools are digital.
Based on the number of the processed images:
- Single-image photogrammetry – It creates material suitable for measurement and mapping from a recording. Only suitable for defining two-dimensional planar data.
- Stereophotogrammetry – Spatial model based on photo pairs.
- Multi-image photogrammetry – Meaningly, modeling of many, many images. It is currently the most common variety. For larger projects, mapping and modeling is based on thousands of photos.
Based on the distance of the camera, the carrier device and the surveyed object:
- Ground photogrammetry – We can talk about ground photogrammetry when photographing from the ground. Based on the relative position of the camera and the captured object, we can further divide:
- Close photogrammetry – The subject distance is within 300 m
- Micro photogrammetry – The scale of the photogrammetric model is greater than 1: 1.
- Mobile photogrammetry – The data collection is made on a moving vehicle
- Aerial photogrammetry – The application for aerial photogrammetry was developed in 1923. Images taken from the air can be categorized according to the angle of the camera axis:
- Vertical – Vertical camera angle (Nadir photography).
- Low oblique – Camera tilt between 3 ° and 30 °. These photos can be useful for taking close-up shots that require more detail.
- High oblique – Camera tilt between 30 ° and 60 °. In this case, it is also possible to map completely vertical geometries (e.g., house walls).
- Space photogrammetry – Files processed from satellite images.
THE OUTPUTS OF PHOTOGRAMMETRY
As I mentioned in part, the following datasets can be generated and updated based on images by using photogrammetry:
- Simple 2D and 3D points, lines, surfaces
- 3D models
- Point clouds – A set of spatial points that represent a specific object or terrain
- Orthophotos – The orthophoto looks similar to a traditional photograph, but has a uniform aspect ratio without any geometric distortion, so accurate measurements can be made on it
- Digital Elevation Models (DEM) – this includes the digital surface model (DSM), which also shows the canopy level for a forest area, and the digital terrain model (DTM), which only represents the soil level.
THE FIELDS OF APPLICATION OF PHOTOGRAMMETRY
Architecture and heritage protection
Probably the most common application field of photogrammetry. Within the category, practical uses such as building facade surveys, motion analysis and deformation measurements, or just documenting historic buildings can be mentioned. It also includes various types of condition surveys, remodeling, pre-design surveys, building damage surveys and interior surveys. It is also a great help in creating three-dimensional city models, which form the basis of several Smart City projects.
Sidenote: If you are interested in how to perform a building motion analysis, read this article as well.
Files generated with photogrammetry can help create forestry maps. For example, an orthophoto and terrain model is suitable for mapping and re-surveying local and regional forest areas. Areas that are difficult to access can also be mapped, and damaged (e.g. wood-beetle) areas can be surveyed thanks to mass data collection. Tree height can also be measured by creating the surface model (DSM) and the terrain model (DTM) of the area, and then determining the tree height of the forest based on the difference. Photogrammetry is also suitable for several other surveys in the field of forestry, some of which are: canopy peak height and diameter, foliage volume, tree trunk circumference and diameter.
Photogrammetry also allows the mapping of different plant coverings, open water surfaces, and soils with different moisture contents. In addition, thanks to aerial photogrammetry, the success of sowing can be quantified on a map by plant counting. The hatched weeds can also be examined for the entire sowing field by generating a vectorized weed map. These data not only provide visual information, but also help in the most efficient spatial distribution of nutrients and fertilizers in precision agriculture.
It is especially true for aerial photogrammetry that very detailed maps and 3D models of very large areas can be made. Maps created in this way contain much more detailed information (topography, vegetation, buildings, hydrography, infrastructure, etc.) than traditional vector maps. It is a great help for local governments, for example, to make it easier to plan and control the infrastructure of the settlement based on high-precision, fast-repeatable recordings.
Photogrammetry is also used in the field of machinery manufacturing. Here it plays a role in the design of machine tools, the calibration of robots used in production, or the recording and analysis of crash tests.
In addition to industrial use, vehicle auxiliaries are already able to use photogrammetry on their own. Some mining machines may also include a photogrammetric data acquisition and analysis system that can automatically calculate the amount of material in a grab bucket or dump truck platform. We can also think of self-driving cars, which with their own monitoring systems are able to analyze road conditions in real time. It should be mentioned here that the spread of LiDAR-based technology is increasingly emerging in the field of self-driving cars.
Photogrammetric modeling and mapping are now standard methods for mapping archaeological research and documenting excavations and finds. It provides designers and researchers with instantly measurable data by applying a variety of maps and 3D models to excavations in urban environments, buildings, or ruined areas. Photogrammetry can be used to extract new data with vector graphic documentation and textured models that more accurately reflect reality.
Medicine and physiology
Photogrammetry is most commonly used in the preparation of plastic and dental procedures and in musculoskeletal analyzes. For example, there are several practice-oriented researches that produce special orthopedic shoes and fasteners with 3D printers that are designed entirely for the feet of patients. The model used for 3D printing is usually based on a photogrammetric model of the patient’s foot. Special uses of the technology can also be made, such as 3D reconstruction of the vascular network of a human heart or brain X-rays.
In the field of natural sciences, the method is mainly used to detect changes over time, for example in the research of raw materials, fluid flow studies, the movement of soils and even glaciers.
Creation of visual effects
This is maybe the most dynamically evolving field of application of photogrammetry, it plays a crucial role in the production of video games and motion pictures. From the implementation of everyday objects in virtual space, the modeling of actors’ movements, to the development of VR (Virtual Reality) and AR (Augmented Reality) technologies, we can find it everywhere.
WHAT WILL BE THE NEXT QUICK START ARTICLE ABOUT?
In the next Quick Start I’ll show you how to take photos with your cell phone for close photogrammetry modeling. Then I’ll present a 3D model creation with freeware programs. I will write these two phases in 2 articles, so there will be data collection and data processing.
In addition, I plan to write a quick-start article on drone photogrammetric route planning and then on maps and 3D models that can be made from aerial photographs. This also means 2 more articles. 🙂
If there was new and useful information in this article as well, please fill out this 5-minute questionnaire so that SurveyTransfer can see the light of day as much as possible tailored to your needs! If you really liked what you read, you can share it with your friends! 🙂
Did you like what you read? Do you want to read similar ones?