Digital Image Interpolation in Matlab

4 Nonadaptive Interpolation

There are a number of treatments for the interpolation process in literature. In this book, we adopt the oldest and most widely accepted definition for interpolation that you can find in modern science:

“to insert (an intermediate term) into a series by estimating or calculating it from surrounding known values”

Such a definition of “interpolation” can be found in the Oxford dictionary and can be fulfilled using model‐based signal processing technique. As a result, in digital signal processing, interpolation is also referred to as model‐based recovery of continuous data from discrete samples within a given range of abscissa [61 ]. In the context of digital image, interpolation is further refined to describe the process that estimates the grayscale values of pixels in a particular high density sampling grid with a given set of pixels at a less dense grid, such that the interpolated image is close to the actual analog image sampled with the high density grid. We have discussed the meaning of having two images being “close,” where the closeness can be measured by any objective and subjective metrics as described in Chapter . An example of image interpolation method known as nearest neighbor has been presented in Section 2.7.3, which follows the algorithm shown in Figure 4.1 to generate an image with twice the size (sampling grid with twice the density) as that of the original image. The size expansion is achieved by first increasing the sampling ...