Digitizing and reconstructing images

Abstract

Before an analogue image can be processed by computer it must first be digitized, or expressed as an array of numbers. The digitizing process can be divided into three phases:

• a) The image area is scanned by an aperture or spot. By definition this is finite in size, and simultaneously acts as a low-pass prefilter (attenuating the higher spatial frequencies). As often as not scanning is of the X — Y type, whereby the image is described line by line. Thus in figure V.1 a small square aperture of side a is being moved across lines which are a apart.

• b) Spatial sampling takes place in two-dimensions. It means accepting from the aperture only those values which describe a series of points forming the nodes of a network. This is usually an orthogonal grid with the same interval gauge in both x and y.

  In the above example this means that an interval equal to the side of the square aperture divides the image into a mosaic of small adjoining squares (figure V.2). Each square is called a picture element or “pixel”. It is assigned a value equal to the aperture response for the position concerned. On this principle figure V.3 shows a sampled image with adjoining square pixels, and illustrates the effect of decreasing the number of elements.