What is Change Detection?
Definition
Change detection is a method that finds differences in a surface or object by comparing observations made at different times. In remote sensing the observations are multi-temporal images of the same area.
How it works
The method has two stages. First the analyst prepares the images. Radiometric correction and image registration make the images from each date comparable. Then a detection algorithm finds the differences.
The input must be consistent. Images from different dates should have the same spatial and radiometric resolution. If the registration is poor, a pixel-based method reports false changes. Images taken near the same date each year reduce the effect of sun angle.
Point clouds
Change detection also works on 3D point clouds. The M3C2 algorithm (Lague and others, 2013) compares two point clouds directly. It measures the change along the local surface normal and gives a confidence interval. Its authors report levels of detection down to 6 mm in situ over ranges of 50 m. That result is for terrestrial laser scanning of a river canyon.
Airfield pavement
An airport must compare pavement condition over time. FAA AC 150/5380-7B says that tracking pavement condition needs an objective, systematic and repeatable evaluation. A federally obligated US airport must make a detailed pavement inspection at least once a year. If it makes a PCI survey to ASTM D5340, the interval may be extended to three years. ICAO Annex 14 requires regular inspection and monitoring of movement area surfaces. Neither document mentions change detection, and neither sets a method for it.
Frequently Asked Questions
- What is change detection?
- Change detection is the comparison of observations of the same area or object made at different times, to find what changed. In remote sensing it means a joint analysis of multi-temporal images of the same area. Applications include land-cover, forest and urban change.
- Why do the images need registration?
- Registration aligns the images from each date so that a pixel covers the same ground point in all of them. Without it, a pixel-based method reports false changes. The images should also have the same spatial and radiometric resolution. Feature-based methods may not need registration.
- What is M3C2?
- M3C2 is an algorithm that compares two point clouds directly in 3D. Lague, Brodu and Leroux published it in 2013. It estimates surface normals at a scale that matches the local roughness and gives a confidence interval for the change. The authors tested it with terrestrial laser scanner data of a river canyon, not with airfield data.
- How often must an airport inspect its pavements?
- In the United States, a federally obligated airport must inspect its airfield pavements in detail at least once a year. If it makes a PCI survey to ASTM D5340, the interval for the detailed inspections may be extended to three years. This is an FAA rule for US airports. It does not name change detection.
References
- khurana-saxena-2015
- Soft Computing Techniques for Change Detection in Remotely Sensed Images: A Review IJCSI International Journal of Computer Science Issues, vol. 12, issue 2 (Khurana and Saxena, 2015)
- lague-2013-m3c2
- Accurate 3D comparison of complex topography with terrestrial laser scanner: application to the Rangitikei canyon (N-Z) Lague, Brodu and Leroux. ISPRS Journal of Photogrammetry and Remote Sensing 82 (2013), arXiv:1302.1183
- faa-ac-150-5380-7b
- AC 150/5380-7B, Airport Pavement Management Program (PMP) FAA
- icao-annex-14-vol-i
- Annex 14 to the Convention on International Civil Aviation, Volume I, 9th edition ICAO (copy hosted by Swiss FOCA)
This is a reference entry. TarmacView provides drone inspection services and doesn't sell the equipment or materials described in this glossary.
