8. Image Processing Part 2

Unit 8 continues to look at the techniques for digital image processing, particularly at the ways that an image can be pre-processed before being visually interpreted or analysed further.   In 8.2, we begin by looking at the ways that we can correct for atmospheric effects in remotely sensed imagery. As we saw earlier in Unit 3, the atmosphere has a profound effect on remotely sensed imagery as a result of processes such as atmospheric scattering, absorption and refraction. In combining images for different parts of the same study area or images from different times and in making measurements from remotely sensed imagery, we often need to take account of these atmospheric effects as the first step in any analysis.  8.3 looks at the ways that we can undertake geometric correction  of remotely sensed imagery. Geometric correction enables locations within an image to be recorded in a geographic reference system, so that – for example – we can find the latitude and longitude of a particular point within an image.  This is particularly important where we wish to link remotely sensed data to other digital map layers held within a GIS.

In 8.4 and 8.5, we look at some image enhancement techniques, particularly image filtering. Image filtering involves enhancing some components of an image, such as patterns of local or regional variation. 8.4 looks at what are known as low-pass filters, which help enhance regional variation within an image. Low-pass filters are often used to remove speckle or noise within an image before further use.  Similarly, 8.5 looks at high-pass features, which enhance local variation within an image. High-pass filters are often used to detect edges, which are areas of sudden change within an image. Edges might include features like roads or streams, but can also be the boundary between different land covers, such as forest-agriculture margins.  Thus, filters provide a way of increasing our ability to draw out particular types of information from within an image.

Activity