IR-Cut Off Filter – A Powerful Weapon for Optimizing Color Imaging

Last week, Matthias posted on the topic of image sensor color reproduction. Today, I’ll be speaking from the camera perspective and offering some insight on how to achieve optimal color quality images in any environment. You may be surprised at how much control you, in fact, have.

Capturing a high-quality color image from a digital video camera requires some knowledge of how a color image is achieved and how external factors can affect performance. Most color cameras are calibrated for white balance and color quality before they ship from the factory but operating temperature can affect white balancing, and the user may have to re-calibrate on-site to accommodate for environmental conditions – temperature and light.

Most of today’s digital, color cameras achieve their color through a “Bayer pattern” of pixels; using dyes that pass a certain range of wavelength, representing the three primary transmitted colors of red, green and blue. These patterns of red, green and blue sensitive pixels are usually arranged in a 2×2 matrix like this:



Even though there may be some overlap in the wavelengths transmitted by each dye (see Fig. 2),  the basic idea is that each colored dye does not trespass too greatly along a wave length other than its own color.

While the peak sensitivity of these pixels is centered around a wavelength that represents the color to be acquired by that pixel, most of these dyes will also be sensitive at longer wavelengths, beyond the visible spectrum (near IR).  Not only are the dyed pixels sensitive to near IR wavelengths, they also tend to be equally sensitive, regardless of whether the pixel has a red, green or blue dye applied.

Since all the pixels are equally sensitive to near IR light, any near IR light, such as one component of sunlight, will tend to wash-out the color differences, making the image appear black and white. This is why it is important, especially in outdoor applications affected by sunlight, to have an IR cutoff filter. This filter, which is typically placed somewhere in the optical path, such as inside the lens mount, will block any wavelengths of light above a certain frequency. A good general purpose IR cutoff frequency is 750 nm.  The figure below illustrates the wavelength transmission for a MidWest Optical 750 nm cutoff filter:

You can see that at around 725 nm the light transmission begins to drop and by the time the wavelength reaches 750 nm, there is virtually no light transmission through the filter.  Also note with this particular filter that the UV wave lengths on the other end of the spectrum are blocked as well.

Choosing the right cutoff filter for your color image application is an essential part of achieving optimal color performance. The good news is the power is in your hands.

Til next time,



About Glen

Glen is Sales & Applications Support Manager for the Eastern US. Glen has been in the vision industry throughout his whole career.
Posted on by Glen. This entry was posted in Cameras, Color imaging, Machine Vision and tagged . Bookmark the permalink.

3 Responses to "IR-Cut Off Filter – A Powerful Weapon for Optimizing Color Imaging"

  1. Yu Xie says:

    pretty cool, but how about the reflectance?

  2. Glen Ahearn says:

    Hi Yu Xie,
    Thank you for your comment. Are you primarily concerned with the reflectance coming into the IR cut off filter – thereby attenuating the total signal into the sensor, or are you concerned with the reflectance in-between the IR cut off filter and the lens (or between the IR cut off filter and the sensor) whereby light could reflect off the “inside” of the cut off filter and be directed back down to the sensor? The asnwer to this question will not change my comment, I am just curious as to where you see most of the reflectance problem coming from.

    In the case of this article, my observations about the IR cut off filter improving color fidelity are primarily directed at application areas where reflectance is not going to pose a major problem. Many of the applications for color imaging still tend to be less precise than their monochrome counterparts. These color applications maybe checking things such as color consistency in printing, or identifying the color of an automobile in a traffic enforcement systems. These applications will typically not be affected by small differences caused by reflection.

  3. James says:

    Hi Glen,
    Do you have the data for blue, green and red in the Relative Response vs. Wavelength graph?