Introduction

What is it for?

Most of the time, when taking a picture of a subject with a digital camera a lot of settings like focus, exposure, aperture, film speed equivalent (ISO) and white balance are determined by the camera based on some simple measurements of the scene. This is fine, where it not that slight changes in that scene, e.g. lighting, can result in quite different images. Manual adjustments of the camera settings usually cannot compensate for this either, and consequently normal digital photography can hardly be called 'reproducible'.

Fig. 1: Consecutive series of pictures taken with a Canon G2 in full automatic mode, but with different lighting: cool white fluorescent, warm white fluorescent and incandescent.

We have developed a software calibration algorithm that effectively compensates for scene and camera setting changes and ensures reproducible imaging of the subject of a scene. It is based on the inclusion of a readily available color chart (MacBeth Color Checker Chart or MBCCC) in the scene, and it delivers images which are defined in a standardized RGB color space called sRGB. This image represents the scene as would look under a blue noon sky, but it can be transformed to other types of lighting. They will display realistically on most modern computer monitors set to 'sRGB' or '6500K' color setting. On top of that the algorithm will also determine the exact physical resolution of the image for the plane of the color chart, so that geometrical measurements (area, length) are possible.

Fig. 2: Consecutive series of pictures with the MacBeth Color Chart taken with a Canon G2 in full automatic mode, but with different lighting: daylight, cool white fluorescent, warm white fluorescent and incandescent. The second row contains the calibrated images. Note that for the incandescent and warm white fluorescent cases the images showed marked saturation in the red channel for red, yellow, and orange color patches. This explains the poor calibration results for the red bolt in the image, and again shows the shortcomings of automatic white balance in some cameras.

Who is it for?

This technology can be of benefit to anyone who needs reproducible imaging, e.g. for comparison over time of a subject. Such a comparison can be qualitative (visual), or quantitiative as this technology also allows precise color measurements to be made on arbitrary regions in an image (sRGB has a fixed and known relationship with the CIE colorimetric color spaces like CIE XYZ, CIE L*a*b*). Because of the realistic display of sRGB images it may also be of use to people who just need consistent display, e.g. of products on a website. Finally, and more fundamentally, because calibration properly defines the meaning of the data in an image this technology allows proper and meaningful exchange of images, be it just for visual inspection, measurements or further image processing. Examples of potential users are:

  • Physicians, for clinical imaging e.g. in dermatology where comparison (both visual and based on measurements derived from those images) over time is wanted for follow-up.
  • Law-enforcement, e.g. in forensic imaging
  • Industrialists, e.g. for product imaging on websites, paper catalogs, databases. Products of intrest could be antiques, interior decoration, exterior decoration (bricks, sidings, ...), art (paintings, sculptures), flowers, ...
  • ...