In any digital imaging system, whether it is a confocal microscope forming an image with a photomultiplier tube (PMT, Chapter 5) or a wide-field microscope using a CCD camera, we are always dealing with an image made up of individual points: pixels. Often we are also handling samples of a three-dimensional volume: voxels. This quantization has profound effects on our image and how we must treat it. Furthermore, each point (unlike a point in a photograph) can have only certain discrete values. In many cases, the number corresponding to one pixel is an
, meaning that it is encoded by eight binary digits (0s and 1s) and can therefore have one of 2
(256) values; it must lie between 0 and 255. Some confocal microscopes, and most scientific CCD cameras, allow 12-or 16-
bit image collection, so that each pixel can have 4,096 or 65,536 possible values. This may seem like overkill, because the eye can perceive only 64 or so shades of gray, but it is not really overkill, particularly if we want to do anything numerical, such as ratiometric measurement, with our image. Dividing one eight-bit value by another does not give us a very wide range of possible gray values. Another situation in which we find ourselves running low on gray scales is when we need to take several images for comparative purposes at the same gain and laser settings; the darkest ones will probably have very few tones if we use only eight bits. Once the number of gray values drops to the point at which the eye can recognize them as separate, the effect is dramatic. Our eyes and our visual processing system have evolved to be very efficient at spotting edges, so our built-in edge detection mechanisms kick in, and all illusion of continuous tone is lost (Figure 6.1). Often the image becomes virtually impossible to interpret. This effect is termed
because it resembles the effect of a poster printed with a limited color palette (historically, because of technical limitations in wood-block printing, but now done for dramatic effect).