Camera vs. Scanner

Digital cameras has increased in resolution dramatically over the past few years. Currently, Canon's lowest price point-and-shoot camera for about $150 has a 16.0 megapixel sensor. Search for the PowerShot SX160 IS. Canon's top-of-the-line full-frame sensor camera has 23.4 megapixels and sells for about $3500 for the body only. The gap between the low end cameras and the high end is closing. If that were the only consideration, then why pay more for slightly higher resolution.

In the Nikon camp, the changes in resolution are even more dramatic.  Nikon's Coolpix, S01 is about $150 dollars and has 10.34 megapixel sensor. At the high end, they have the Nikon D800 with a 36.3 megapixel full-frame sensor for around $3000 for the camera body alone.

The question is whether or not this increase in relatively low-cost digital camera resolution has now reached the point of replacing the traditional flatbed scanner? Or, is possibly the case, only the high-end cameras have the quality to compete with flatbed scanners?

Scanners are often advertised using an inappropriate measure of resolution, that is, dots per inch or DPI. The reason it is inappropriate is rather simple, dots per inch refers to printing, i.e. a laser printer puts 300 dpi of toner on a piece of paper. Scanners use a bank of CCDs (charge-coupled devices) to capture the image. A CCD is a collection of tiny light-sensitive diodes, which convert photons (light) into electrons (electrical charge). There are other technologies, including the CIS (Contact Image Sensor), and the PMT (photomultiplier tube) which are found, respectively, in the low and high ends of the scanner market. No matter which type of sensor is used, the scanning mechanism shines a light on the object or document being digitized and through a series of mirrors and/or lenses, projects the image on the sensors.

The key to understanding scanner technology is that the resolution of the scanner can be effectively no greater than the number of sensors in the imaging array and the precision of the stepping motor that moves the array across the object or document to be scanned. Here is a quote from howstuffworks.com that explains the issue:

Let's take a simple example: If a scanner's resolution is 300x300 dpi, and that scanner is capable of scanning a letter-sized (8.5x11-inch) document, then the CCD has 2,550 sensors arranged in each horizontal row -- 8.5 (inches across) x 300 (x-direction sampling rate) = 2,550. A single-pass scanner would have three of these rows for a total of 7,650 sensors. The stepper motor in our example is able to move in increments equal to 1/300ths of an inch.

If you look at the technical specifications for scanners, it is extremely difficult to get an accurate idea of the total number of sensors. Sometimes the total number of sensors is disguised as "optical resolution." Optical resolution is a measure of maximum hardware sampling resolution, based on ISO 14473 standard. For example, the CanoScan 9000F film scanner from Canon, claims an optical resolution of 9600x9600 dpi. Since this a one pass scanner, in order to have a true resolution of 9600 dpi, the scanner will need a bank of three sensors that would have 8.5 (inches across) x 9600 (x-direction sampling rate) or 81,600 sensors. This is really not the case. Despite the characterization of the resolution as "optical" cramming more sensors into the horizontal CCD sensor is not physically possible.

In effect, different arrangements of the scanning sensors and software interpolations give the appearance of a higher resolution. But you can't beat physics. There are limitations to any optical system such as the wave length of the light involved and more importantly, the resolution power of the human eye.

So if DPI aren't realistic, then how do we measure resolution? Simple. Through scanning or photographing a standardized test pattern made up of small groups of parallel lines at different sizes. The target looks something like this:

SilverFast_Resolution_Target_USAF_1951.png http://en.wikipedia.org/wiki/File:SilverFast_Resolution_Target_USAF_1951.

As they say, the proof is in the pudding, the way to determine if either a scanner or a digital camera will effectively produce a better image is to try both methods and compare them carefully at high magnification on the computer screen. As I did this with each digital camera I purchased, the flatbed scanner always produced a superior image. Finally, when the camera got to over 20 megapixels, with a full-size 35mm equivalent sensor, the resolution of the camera exceeded that of the flatbed scanner.

But for most purposes they are of equal quality, especially if you are trying to reproduce documents. The digital camera at about 10 megapixels, matched the flatbed scanner.

If I get some time, I will post some comparisons at high magnifications.