Image Analysis Terms

Image depth:

The number of bytes per pixel. An image with a higher image depth has a greater number of possible intensity values, which improves the quality of filtering and measurement.

Pixel value:

The raw number representing a pixel, as it occurs in an image file. This number can have any number of bits from 1 to 48, corresponding to 2 to 2.8$\times$10$^{14}$ different colors.

Pixel intensity value:

The brightness of a pixel. For grayscale images, this is the same as the pixel value. For measurement purposes, the pixel intensity is scaled to {0..1} so that the results are independent of the image depth. The actual shade of gray displayed on the screen may be different, depending on whether gamma correction is used.

Gamma correction:

A nonlinear correction factor which is supposed to convert an evenly-spaced grayscale palette to evenly-spaced levels of brightness in the monitor (as measured by a photometer). It is usually in the form of a table representing the function

$$output = input ^{gamma}$$

where gamma is usually 1.7-2.2.

Pixel density:

The grayscale intensity of a pixel expressed as a number in the range 0..1.

Luminosity:

The grayscale intensity of a color pixel, computed from

$$luminosity = a\times red + b\times green + c\times blue$$

where $a, b, c$ are in the range 0..1.

Image types

Images fall into 3 distinct categories:

1. Grayscale: Pixel values directly correspond to intensity. There may in addition be a table mapping each intensity to an optical density value. Grayscale images are usually 8, 12, or 16 bits/pixel, for 256, 4096, or 65536 gray levels, respectively. Since the monitor can only display 256 gray levels at a time, these gray levels are reduced to 256 in imal by a grayscale intensity map which can be adjusted to highlight any specific intensity range.

   ``False-color'' images are a subset of grayscale images in which the computer displays an arbitrary color instead of a shade of gray.

2. Indexed-color: Pixel values have no intrinsic meaning and are merely indices into a color map containing 256 colors in a random sequence which is totally different for each image. Indexed-color images are almost always 8 bits/pixel.
3. Color: Pixel values are a composite number containing intensities of red, green, and blue. The number of bits of each color is different for each standard image depth (i.e., 15, 16, 24, 32, and 48 bits/pixel).

Each image type has benefits and drawbacks. For example, cutting and pasting among different indexed-color images requires finding the closest matching colors. If the color maps are too different, the pasted image will not have the same color as the original. This problem can be avoided by converting the images to true color first, then converting back to the original image type afterwards. Similarly, densitometry, wavelets, and FFT only work well on grayscale images. Before performing FFTs, it is necessary to convert color images to grayscale. For indexed-color images, the additional step of sorting the colormap is necessary to ensure that pixel values bear a linear correspondence to intensity. Converting grayscale images to color involves defining or selecting a colormap and converting the image to indexed color, and then converting it to true color. All of these operations are done automatically in imal.

When the mouse is moved in imal, the information area shows the x and y coordinates and the value of the pixel at the mouse position. This is convenient when manipulating the image, since much image manipulation uses simple mathematical functions to transform the image. Thus, to make the image lighter, one could ``add'' a value of 50 to each pixel. To increase contrast, one can multiply each pixel value by a constant factor. In contrast to some programs on some other operating systems, these steps are not hidden from the user merely for the sake of ease of use. This ensures that quantitative manipulations on the image remain directly traceable to the original data.