Image Analysis Using QGIS
Introduction
Rasters are created from gridded data. Each pixel is coloured according to an interpolated value, e.g. triangulation (TIN), nearest neighbour analysis.
A raster file is comprised of a pixels arranged in a grid formation. Each pixel contains a colour value that instructs the computer as to what colour to use when displaying it. Raster images tend to be used for grids as they are a more efficient method of showing large areas of coloured pixels than vector maps.
The following illustrates how a raster grid represents terrain, and how the information might be extracted:-
For simplicity’s sake, imagine that we’re back in the days of 256 colours with 1 being white and 255 being black. I tend to display relief with the highest ground as white or red, then to show lower ground as green or blue.
Let’s take a cross section through a hill: 
A grid raster image of the terrain would appear similar to below (please note that I have drawn this in Inkscape using a gradient fill to keep the demonstration simple!):-
The numeric values of the raster grid that the computer would see would be similar to this:
| 5 | 5 | 5 | 5 | 5 |
| 5 | 100 | 150 | 100 | 5 |
| 5 | 100 | 250 | 100 | 5 |
| 5 | 100 | 150 | 100 | 5 |
| 5 | 5 | 5 | 5 | 5 |
Note the values are not the actual elevation, just the colour values of the pixels. The elevation that each pixel value corresponds to (the legend) is contained in the accompanying shape file along with image registration (the x, y coordinates).
By analysing the grid and determining the relationship between pixel values and the elevation that they represent the GIS software can accurately model the terrain. Once the terrain has been modelled, it is possible to undertake further analysis such as slope calculation, predicting hill shade or water runoff.
The Image toolbar
Firstly, let’s have a look at QGIS’ image tool bar:-
QGIS image bar
This can be added by right clicking on any toolbar and selecting Raster from the short cut menu. The buttons are from left to right:-
- Stretch Histogram to Full Data Set
- Local Histogram Stretch
- Geo-referencer
- Interpolation
- Zonal Statistics
