ISPRS Camera Calibration and Performance Database
This project is funded by the ISPRS Educational and Capacity Building Initiative in 2018 to establish an online database of camera calibration and performance information that will provide a unique and comprehensive source of information for anyone using cameras or camera systems for measurement or mapping.
The intent of the database is to cater to any type of camera used for measurement and mapping, however the potential range of cameras is very broad, so the emphasis is cameras used for close-range (terrestrial), unmanned drone mounted and manned aircraft mounted cameras. The online database contains information on the specifications of camera and lens combinations; the results of camera calibrations in terms of quantification of calibration parameters; levels of precision and accuracy resulting from the calibration or project; and a link to further information on the calibration or project such as reports and papers.
The database will be a public resource that will enable practitioners to both contribute and extract information on cameras. The database will allow anyone to evaluate the suitability of cameras for a project or compare the performance of cameras against their own experience.
No idea what this is all about or how to start? Start by reading the database review and testing guide for the project.
Mark R Shortis - March 2020
Frequently Asked Questions
This FAQ for the camera calibration project is (always) a work in progress. New FAQ sections are being added regularly.
What are the levels of membership?
There are three levels of membership, guest, member and editor. A guest member can search the database and use the utilities. A registered member can also create and edit their own calibration data examples to contribute to the database. An editor has full access to all records in the database and can also, for example, add new entries for camera and lens manufacturers.
Do I need to register?
If you only want to search the database and use the utilities, no, there is no need to register. However if you want to contribute to the database with your own examples of calibration data then you must register as a member. Member information is used to associate specific calibration records with the author.
Camera and Lens Information
What are the camera types?
Most of the different types should be self-explanatory. DSLR refers to Digital Single Lens Reflex, essentially a digital version of the traditional 35mm film cameras. A compact DSLR has similar functionality with a smaller physical size. GigE cameras transmit images over a gigabit Ethernet link. Scientific cameras are typically large format, high sensitivity, high resolution sensors with 35mm SLR style optics.
How do I find out what type of sensor is in my camera?
The vast majority of recent digital cameras use a Complimentary Metal-Oxide Semi-conductor (CMOS) sensor. Older digital cameras, typically pre-2000, use Charge-Coupled Device (CCD) sensors. ToF is a Time-of-Flight camera that measures both intensity and range. The sensor type should be available in the specifications from the manufacturer's web site or from Digital Photography Review
What is the Sensor Resolution Spacing?
This is the physical size or spacing of each sensitive element on the sensor. Effectively this is the size of the pixels in the image. Compact and phone cameras may have pixels as small as 1 micron, while scientific cameras may have pixels as large as 25 microns. This information may be available in the specifications from the camera manufacturer's web site or from Digital Photography Review
How do I find out the type of fisheye lens used by my camera?
Most modern fisheye lenses use the equidistant or equisolid angle mapping function. For more information on fisheye lenses see the Wikipedia article. Information on the mapping function should be available in the specifications from the lens manufacturer's web site.
Are the calibration parameters the value or a correction?
All calibration parameter values represent the actual, physical distortion, not the correction (also known as the error) of the distortion. For example, if a lens exhibits barrel distortion then the radial lens parameter would be a negative value.
What are the units of the calibration parameters?
In general the units of the calibration parameters are in millimetres (mm). The units of the lens distortion parameters are adopted such that the multiplication of the parameter by the radial distance raised to the appropriate power gives a result in mm. For example parameter K3 (mm/mm3) multiplied by the cube of the radial distance (mm3) results in a distortion value in mm. Some values must be provided in micrometres (um, also known as microns) or pixels.
Are the calibration parameters based on a particular model?
In general the calibration model is based on the work of Duane Brown. See his papers:
Brown, D. C., 1971. Close-range camera calibration. Photogrammetric Engineering and Remote Sensing 37(8): 855-866.
Brown, D. C., 1966. Decentring distortion of lenses. Photogrammetric Engineering 22: 444-462.
The radial distortion function of odd-ordered polynomials is documented in:
Ziemann, H. and El-Hakim, S. F., 1983. On the definition of lens distortion reference data with odd-powered polynomials. Canadian Surveyor 37(3): 135-143.
What is calibration stability?
Calibration stability refers to the camera(s). Are all of the calibration parameters stable during the calibration photography? Any calibration parameter that is likely to be the same should be applied equally to every exposure (ie photograph) taken by the camera. Any calibration parameter that is likely to change should be applied differently to every exposure. Block-invariant means that the same calibration parameters are used for every exposure from a camera. For example, for a stable lens the radial distortion should be consistent for all photographs, so these parameters are considered to be block-invariant. Photo-invariant means that some or all of the calibration parameters are different for every exposure for a camera (because the camera, lens or optical path is not stable). For example, the principal point location will not be stable for a camera with a body or lens mount that is not rigid, so these parameters could be considered to be photo-invariant.
For more information see:
Shortis, M. R., Robson, S. and Beyer, H. A., 1998. Principal point behaviour and calibration parameter models for Kodak DCS cameras. Photogrammetric Record, 16(92): 165-186.
What are typical calibration objects?
The most convenient calibration object is a flat plane, such as a checkerboard or a flat plate. This is a 2D object. Another common approach is a 2.5D object, which is non-planar but typically with one dimension smaller than the others. An example is a flat plate with offset targets. A cubic structure or a purpose-built 3D test range has similar dimensions in all three coordinates. This type of 3D object is more complex but, all other things being equal, provides the most reliable calibration.
What should I upload as a calibration document?
The calibration document is typically a description of the calibration process and outcomes. It will most likely be a published paper or an unpublished report, a calibration report generated by the software, or a calibration certificate provided by a vendor. Any document that provides detailed, relevant information would be a valuable contribution.
What should I provide in the reference information?
If the calibration has been described in a published paper or report provided as the calibration document, provide the reference here eg Smith A. and Jones, B., 2000. A new camera calibration technique. Journal of Camera Performance, 1(1):1-10.
What should I provide in the additional information?
Anything that is helpful and is not provided in other documentation. For example, exposure details (eg the shutter speed, ISO, f-stop) are often useful. Or, if you have selected ‘other’ in any of the fields for the calibration data, you can provide a brief explanation here.
What should I provide in the online information?
If the paper or report you have noted has been published online, enter a valid web address (eg https://www.mdpi.com/1424-8220/15/12/29831)