Bayer filter array
Alright, so last week I wrote about your camera’s sensor technologies. Today, I cover another important sensor topic, color. Natively, your image sensor is “color blind”, it can only “see” grayscale. To remedy this, the sensor has to be fashioned with a pair of shades. Let’s call it a filter.
In my previous post I described each photosite as a well or cavity. In order to capture color, each cavity is covered with either a red, blue, or green filter, allowing each photo-receptor to record only one of the three colors.
The most commonly used filter is called a “Bayer filter array”, you might hear it referred to as a “Bayer filter mosaic”. As you can see by the example, it is an alternating blue/green and green/red pattern. You might also notice that there are more green squares than blue or red. In fact the Bayer filter is 50% green, 25% blue, and 25% red. In an effort to mimic the human eye, this was done because the eye is more sensitive to green light than red and blue.
How does a complete image get produced if each photosite can only read one color?
This is where the guesswork comes to play. Since each photosite can only capture one color, 2/3′s of the light/color is lost. To make up for the loss, a process called demosaicing is used. Each photosite uses the data from adjacent photosites, then the data is run through a number of algorithms (problem solving instructions). Finally, it gets translated and estimated into a color value containing different levels of red, blue, and green. That process is called interpolation.
In all fairness there are variations on the Bayer pattern. Some sensors use the complementary colors (cyan, magenta, and yellow). The Sony sensor uses two different shades of green. There is also another interesting technology, it’s called the Foveon sensor, owned by the Sigma corporation. You might know them by the lenses they manufacture.
Instead of the lateral color arrangement of the Bayer, the Foveon uses a vertical approach. Using three levels, the color filters and photo-receptors are stacked allowing each photosite to record all three colors simultaneously (three photo-receptors for each photosite). This construction is similar to the way color film is made using a layer of emulsion for each of the three colors. This process eliminates the need for demosaicing and interpolation, which creates faster processing of images and possible eradication of artifacts created from demosaicing. The manufacturer also claims lower noise, improved low-light performance, higher dynamic range, and richer color tones.
That’s it for today, thanks for reading. Good luck and remember to post any questions or comments. I’ll catch you on the next Techie Tuesday.
Peter
Peter Pechacek is a photographer and filmmaker in Orlando, Florida. He is new to the blog and will be contributing weekly on Techie Tuesday.
