Improve Color Quality Using Color Calibration

If you look at a scene containing color objects these objects should appear pretty constant whatever the light source is used to illuminate that scene. But when you look at the image coming out of your color camera shooting that same scene you notice that colors vary – e.g. it may appear slightly bluish or reddish – depending on what type of light is chosen (incandescent, halogen, fluorescent, etc). In fact our eyes and our brain perform a chromatic adaptation in order to keep colors independent of light; which is called color constancy.

Color Perception by Human and Device

Fortunately color calibration exists to help devices behave like humans! It is a technique by which a sensor is trained under different light sources to provide color constancy. Color calibration is based on a device-independent color space (e.g. CIE XYZ) which represents the full visible color range. Its purpose is to characterize a sensor, i.e. to find a model which describes the response of a particular sensor in respect to all visible colors.

Color calibration is performed in two main steps. First the characterization consists of finding the transformation between the device color space and an independent color space (XYZ) with respect to light. It is done only once per sensor and typically performed in a lab under strictly controlled lighting. Second, the correction is the computation of the final colors (corresponding to a standard color space such as sRGB) through a mathematical formula including the characterization coefficients. The correction is applied on every incoming image and is usually embedded in the camera hardware to permit real-time operation.

To perform characterization, you don’t need all the visible colors. Only a set of a few well chosen colors are sufficient. In practice the well-known 24-patch Gretag-Macbeth color chart is perfect.

Gretag-Macbeth Color Chart

First, a raw color image of the color chart is captured from which an algorithm computes the average value on each patch of the chart. Then a second algorithm computes the transformation from the measured values to the standardized (final) values based on the type of light used to illuminate the chart.

Each type of light has its own color temperature which characterizes its spectrum. Color temperature can be measured using a spectrometer (on a white source) or simply selected from a list of theoretical light sources. Common sources are incandescent (tungsten or halogen), daylight and fluorescent, and their corresponding temperature typically ranges from 2800K to 4000K. Finally a gamma factor is optionally applied to the final values in order to bring the level of intensity to be visually similar to your display monitor. Here is an example of what a color chart looks like before (left) and after (right) color correction.

Effect of Color Correction

That was color calibration at a glance. Today’s color cameras provide integrated color correction that helps you get the best colors out of your scene of interest.

About Bruno

Bruno is group leader of image processing at Teledyne DALSA. He holds over 16 years of experience in image processing research and development and software design. He has a Master’s degree of Systems Engineering specializing in computer vision and expert systems, as well as a Bachelor’s degree in Electrical Engineering, both from École de Technologie Supérieure in Montreal. He manages the development of image processing algorithms, libraries and tools, both at the software and hardware levels.
Posted on by Bruno. This entry was posted in Cameras, Image processing, Machine Vision and tagged , , , . Bookmark the permalink.

Comments are closed.