Digitization of Gray Scale Images

When a picture is to be processed by computer, it is often described as a matrix, or some other discrete data structure. But a picture is primarily a signal that conveys information to an observer, and there are many applications where this consideration is particularly important. We shall devote this and the next chapter to the discussion of such problems, especially for gray scale (class 1) images. The first problem is the conversion of a continuous picture into a discrete form, and this involves two processes: sampling, which is the selection of a discrete grid to represent an image, and quantization, which is the mapping of the brightness and color values into integers. In graphics one is concerned with similar problems: specifically the choice of the display resolution and number of gray levels or colors. These processes are also relevant in one-dimensional data and have been studied thoroughly in that case, but two-dimensional data present new problems. We shall devote the first part of this chapter to a review of transform techniques and then we shall discuss sampling for the one-dimensional case, followed by sampling for pictures. Quantization will be treated in the last section.