Date: 05 Apr 2013

Autocorrelation Function

* Final gross prices may vary according to local VAT.

Get Access

Abstract

In this chapter, as in the previous one, we look at images directly—we do not try to segment the images, we do not look for objects that are represented on the image. Instead, we will again take the gray values for what they are, and we will think of the image matrix as a map of measurements that are coded as gray values. We will then look at the statistics of these measurements, and we will be concerned with describing the spatial correlation between them. The two-dimensional (2-D) autocorrelation function is an ideal tool for describing such maps or ‘visual textures’. As will become apparent, the calculation of the autocorrelation (ACF) is easy and it is fast because no prior segmentation is necessary. The difficult part of the ACF analysis is to interpret the ACF and relate its geometrical characteristics to the image for which it was calculated (Heilbronner 1992).