Most of the sensors currently in use are based on CCD technology. Although this technology is quite advanced, actual CCD production is highly specialized. As a result, CCD sensors are expensive. The production of CMOS sensors is based on the same technology used to produce microprocessors and other computer chips. As a result, their production cost is less. Also, additional circuitry can be added on the same chip as the sensor, eliminating the need for external processing chips required with CCD sensors. A problem with early CMOS sensors was the high background noise. However, that problem has been solved. The highly successful Canon D-30 and D-60 SLR cameras use CMOS sensors and produce very little background noise. In time, CMOS sensors will catch up and pass CCD sensors as the sensor of choice.
Did you know that the sensor photosites are color blind? Each photosite simply collects the amount of light striking it and converts it to electrical charge. There is no color data generated. So, how is color data generated? There are several ways to do this. You could split the light coming through the camera lens and route it to three different sensors, each of which could be adjusted to react to a certain portion of the light spectrum. This is done by covering each sensor with a filter colored to accept one of the three primary colors: red, green, or blue. This technique is used in top-of-the-line cameras.
However, the most common technique, which is used in most consumer cameras, is to place an array of colored filters over the photosites in a single sensor. Two filter arrays are commonly used: RGGB and CYMG. With the RGGB pattern, odd numbered rows of photosites are covered by alternating red and green filters. Even numbered rows are covered by alternating green and blue filters. The RGGB pattern is commonly referred to as the Bayer pattern after the Kodak engineer that invented it. This array uses the primary colors (red, green, and blue) of the additive mixing method used in television and computer monitors and is the pattern used in most digital cameras.
The other array, CYMG, is a more complex filter array. It uses the primary colors (cyan, yellow, and magenta) in the subtractive process (commonly used in printers) plus green. This array is used in some of the Nikon Coolpix series of cameras. Since the eye is most sensitive to the color green, green filters are added to this array. Likewise, in the RGGB array, there are twice as many green filters as red or blue.
So, how do digital cameras achieve their advertised resolution if it takes four separate pixels to determine the color of a single pixel. The answer is interpolation. The camera's software examines a block of adjacent pixels to determine the missing values for each pixel location. For example, consider a red pixel. The software will determine its blue value by averaging the values of the surrounding blue pixels. Likewise, it will determine the green value by averaging the surrounding green pixels. The software used to do this is commonly called demosaicing algorithms, because it converts a mosaic of separate colors into a mosaic of true colors.
Another somewhat confusing parameter associated with camera sensors is the actual physical size of the sensor. In a camera's specifications, one will usually see a fractional designation, such as 1/2.7", 1/1.8", 2/3", or 4/3". Occasionally, you may see the actual physical size, i.e., the length, width, and diagonal, given in mm. These numbers seem to have no relationship to the fractional designation.
The fractional designation originated with a set of standard sizes given to television vidicon tubes in the 1950s. These sizes were given as 1/2", 2/3", etc. These numbers did not define the diagonal of the sensor area, but the outer diameter of the glass envelope of the tube. It was soon discovered that the diagonal of the usable image area was approximately two-thirds that of the designated size. Unfortunately, the original sizes are still used; a much more useful measurement would be the actual diagonal of the sensor. The following table lists some common sensor sizes.
| Type | Aspect Ratio | Sensor Diagonal | Sensor Width | Sensor Height |
| 1/2.7" | 4:3 | 6.59 | 5.27 | 3.96 |
| 1/2" | 4:3 | 8.00 | 6.40 | 4.80 |
| 1/1.8" | 4:3 | 8.93 | 7.18 | 5.32 |
| 2/3" | 4:3 | 11.00 | 8.80 | 6.60 |
| 1" | 4:3 | 16.00 | 12.80 | 9.60 |
| 4/3" | 4:3 | 22.50 | 18.00 | 13.50 |
| 35-mm film | 3:2 | 43.30 | 36.00 | 24.00 |
As to which sensors are being used, the 1- and 2-Mpixel cameras use the type 1/2.7" or type 1/2". Most of the 3- and 4-Mpixel cameras use the 1/1.8" sensors. The 5-Mpixel cameras, such as the Sony DSC-F717 and Nikon 5700, use the 2/3" sensor. The Canon D-60 and Nikon D100 cameras, which are 6-Mpixel cameras, use the 4/3" sensor. Olympus has recently announced a new line of cameras to be released in the fall, based on the 4/3" sensor. In fact, Olympus, Kodak, and Fujifilm are working together to produce a new standard for SLR cameras with removable lenses based on the type 4/3" sensor. Camera bodies and lenses based on this standard should be much less expensive than the current generation of digital SLR cameras using lenses based on 35-mm film technology.
The future of digital cameras is truly exciting. The production of cheaper and larger sensors will soon result in consumer-priced cameras that will actually surpass film cameras in the quality of images produced. For more information on digital camera sensors, visit the following web sites: www.normankoren.com/digital_cameras.html, www.luminous-landscape.com/essays/counting1.shtml, and www.bythom.com/ccds.htm.
T. Zinneman is a CCPCUG member. If you have any questions or comments, please send them to tzinneman@comcast.net.
For more information on the Tulsa Computer Society click here