/photogrammetry

Photogrammetry Introduction, References, and History

Primary LanguageCSS

layout title permalink
default
Intro to Photogrammetry
index.html

Intro to Photogrammetry

Author: Gabriel Dunne
email: gabe@gabrieldunne.com
www: gabrieldunne.com
twitter: @gabrieldunne
ig: @gabrieldunne

Headlands Bunker Scan (Dense Cloud with Image Planes), Agisoft Screenshot. Source: Okaynokay, Twilight Engines 2019

What is Photogrammetry?

Photogrammetry is the process of creating 3D objects from 2D photographs. The process can be used to scan anything from objects and people, to architecture, terrain and landscapes. There are various strategies you can use when taking photos to produce accurate models. This document includes photography techniques, ideal camera settings and lighting, photogrammetry software, and further resources to create 3D models from photos.

Structure from Motion (SfM) photogrammetric principle. Source: theia-sfm.org 2016

Definition

Photogrammetry [noun] The science or technique for obtaining reliable information on the natural environment or physical objects by recording, measuring and interpreting photographic images.

Greek:

  • "photos" (light)
  • "gramma" (something written or drawn)
  • "metron" (measure)

Typical Outputs

A map, a drawing, a 3D model of a real-world object, scene, or terrain.

Related fields

Remote Sensing, GIS, Stereoscopy

Main Tasks of Photogrammetry

  • To measure something without touching it
  • To measure something that may no longer exist, or may only exist in photographs
  • To measure something too large to measure with traditional methods, i.e., landscape, a megalithic structure
  • Quantitative data from photographs, the science of measuring in photos.

Comparison to Laser Scanning (LIDAR)

Laser scanning is great in cases where photogrammetric techniques often fail -- such as when objects have low texture, or shiny/reflective. However, laser scanning is expensive, time consuming, and can't be used with moving objects like Photogrammetry can. That said, Laser Scanning and Photogrammetry can be considered complimentary to eachother.

Abridged Photogrammetry History

Illustration from Oculus artificialis teledioptricus sive Telescopium, Johann Zahn 1685, Source: Public Domain
  • 1046 BC - 256 BC Discovery and capture of natural optical phenomena. Perforated gnomons projecting a pinhole image of the sun were described in the Chinese Zhoubi Suanjing writings. Some ancient sightings of gods and spirits, especially in temple worship, are thought to possibly have been conjured up by means of camera obscura projections.
  • 300 BC Geometry, perspective, pinhole camera model — Euclid
  • 500 BC Camera Obscura. (500 BC in China, 350 BC Aristote, 1000 Al-Haytham, 1500 Léonard de Vinci)
  • 1430 Leon Battista Alberti, "Rerum Mathematicarum et Descriptio Urbis Romae". Developments of topographic mapping.
  • 1611 First documents of Camera Lucida, eventually Patented in 1806 by William Hyde Wollaston.
  • 1836 Daguerreotype, “gift to the world” from French Academy. First publicly available photographic process.
  • 1851 French officer Aime Laussedat develops the first photogrammetrical devices and methods.
  • 1858 The German architect A. Meydenbauer develops photogrammetrical techniques for the documentation of buildings.
  • 1866 The Viennese physicist Ernst Mach publishes the idea to use the stereoscope to estimate volumetric measures.
  • 1885 The ancient ruins of Persepolis were the first archaeological object recorded photogrammetrically.
  • 1889 The first German manual of photogrammetry was published by C. Koppe.
  • 1893 Meydenbauer coined the word “Photogrammetry”.
  • 1896 Eduard Gaston and Daniel Deville present the first stereoscopic instrument for vectorized mapping.
  • 1910 The ISP (International Society for Photogrammetry), now ISPRS, was founded by E. Dolezal in Austria.
  • 1911 Aerial photogrammetry with rectified photographs by Theodor Scheimpflug.
  • 1924 Relative orientation determined by 6 points in overlapping images — von Gruber points.
  • 1957 Analytical plotter (Helava) - Image-map coordinate transformation by electronic computation & servocontrol.
  • 1964 First architectural tests with the new stereometric camera-system, which had been invented by Carl Zeiss, Oberkochen and Hans Foramitti, Vienna.
  • 1980 Due to improvements in computer hardware and software, digital photogrammetry gains more and more accessibility.
  • 2000's Increasing accessibiulity of photography drones

Image Capture Technique

Illustration from Oculus artificialis teledioptricus sive Telescopium, Johann Zahn 1685, Source: Public Domain

Camera

  • The best camera is the one you have that offers the sharpest image
  • RAW format if possible for best quality
  • Large storage if you take a lot of pictures, especially if you shoot in RAW
  • Pixel count is less important than physical optics (lens) quality and clarity.
  • Use a fixed focal length during the whole session. Avoid lens distortion. If you have a physical zoom, make sure it doesn't shift during your session.
  • ISO: As low as possible
  • Shutter speed: As fast as possible
  • Aperture: F8, higher is better
  • 50mm (35mm equiv) Optimal Focal Length
  • White Balance: Manual
  • Bokeh is bad! Increase aperture for larger DOF for sharper photos.
  • Overcast/foggy/diffuse lighting is ideal — Soft shadows & low contrast for evenly lit images.

Optional Equipment

  • Tripod/Monopod for stable shots
  • Color Checker For color accuracy
  • Scale Bar for measurement
  • Ground control points for picture overlap
  • Spray: Water or paint for reflective surfaces
  • Turntable for small objects in a controlled environment)
  • Drone for large swaths of terrain, autonomous aerial scans

Guidelines

Source: Agisoft Photoscan Manual
  • Don’t worry about taking too many photos. Remove bad ones later.
  • Use the highest resolution possible
  • Prioritize a sharp, focused image. Avoid movement, motion blur, or out of focus areas.
  • Use a remote trigger for your camera when possible for an even sharper image.
  • Enable GPS on your camera if supported (improves alignment in software)
  • Each point on the scene must be seen in at least TWO images, and you should aim for THREE. Optimal overlap is about 80%
  • Take a note of a measurement of an element in the scene, or include a ruler or measuring tool somewhere in the scan.
  • Always physically move when taking photos. DO NOT just rotate.
  • Rotate the camera on both axis, not just one
  • Don’t change a rotational viewpoint more than 30 degrees in a series
  • Prioritize object texture.
  • Avoid plain/monotonous or glittering/reflective surfaces.
  • Avoid harsh shadows or high contrast areas. These may be confused with geometry by the alignment software.
  • Avoid shiny, glittery surfaces.
  • Avoid anything with a specular highlight that changes as the camera view moves
  • Start with the entire scene or object, and gradually hone in on details
  • Complete loops. Go around objects and end up where you started
  • Pictures are processed in order, so stay organized
  • Don't just take one loop, take multiples from various angles
  • Turntable Method: Object moves, camera stays still
  • Walk-Around Method: Object stays still, camera moves
  • Go with your instincts, break rules, test ideas!

Practice

Headlands Rock Scan, (Sparse Point Cloud). Agisoft Screenshot. Source: Gabriel Dunne 2019

The best way to learn is to practice. Here are some ideas for things to scan.

  • A rock
  • Statue or a monument
  • Building facade
  • Fire hydrant
  • A curb or urban object
  • A vehicle like a car or a bicycle or skateboard
  • Shoe
  • Handbag or backpack
  • Tunnel or some stairs
  • Room, interior scene
  • Building Exterior
  • A reflective surface, a mirror in a room (will break spectacularly!)
  • Someones head or body (they have to be still!)
  • An Insect
  • A plant
  • Food
  • Terrain with a Drone

Photogrammatry Solutions/Apps/Software

Free 3D Modeling, Mesh Manipulation Software Suggestions

Articles and References

Illustration from Oculus artificialis teledioptricus sive Telescopium, Johann Zahn 1685, Source: Public Domain

These articles are for reference and educational use. I claim no ownership or rights to this material.

Emergence of Mathematical Solutions related to Photogrammetry

© Gabriel Dunne, 2019