Digital interpretation and Image classification

Extraction of Data

detection identification analysis and deduction classification theorization (verify/nullify hypotheses)

Image elements

tone/color least complex size shape texture pattern height shadow association pattern most complex

Computer-assisted classification

classifying raster data automate of low complexity functions preprocessing: radiometric & geometric correction classification approaches

supervised classes assigned unsupervised - cluster analysis hybrid unsupervised followed by supervised

Change detection

overlay

map-to-map image-to-image
matrix map

output

Image classification

In computer-based quantitative analysis the attributes of each pixel (such as the spectral bands available) are examined in order to give the pixel a label identifying it as belonging to a particular class of pixels of interest to the user. As a result, the process is often also called classification.

False colour composite

A colour composite in which selected features or bands in multispectral data are chosen to be associated with the three additive colour primaries in the display device which produces the colour product. When the data consists of more than three features a judgement has to be made as to how to discard all but three, or alternatively, a mapping has to be invented that will allow all the features to be combined suitably into the three primaries.

False colour composite

In large number of bands only experience, and the area of application, tell which three bands should be combined into a colour product. For data with limited spectral bands however the choice is more straightforward. An example of is Landsat Thematic Mapper data. Use FCC of bands 5, 4, 2 (Red, Green, Blue) for better visualization of features in natural colors.

Quantitative Analysis

Computer interpretation of remote sensing image data is referred to as quantitative analysis because of its ability to identify pixels based upon their numerical properties and owing to its ability for counting pixels for area estimates. It is also generally called classification, which is a method by which labels may be attached to pixels in view of their spectral character. This labelling is implemented by a computer by having trained it beforehand to recognise pixels with spectral similarities.

The most effective means by which multispectral data can be represented in order to formulate algorithms for quantitative analysis is to plot them in a pattern space, or multispectral vector space, with as many dimensions as there are spectral components. In this space, each pixel of an image plots as a point with co-ordinates given by the brightness value of the pixel in each component.

Information classes

Provided the spectral bands have been designed to provide good discrimination it is expected that pixels would form groups in multispectral space corresponding to various ground cover types, the sizes and shapes of the groups being dependent upon varieties of cover type, systematic noise and topographic effects. The groups or clusters of pixel points are referred to as information classes since they are the actual classes of data which a computer will need to be able to recognise.

Spectral classes

In practice the information class groupings may not be single clusters. Instead it is not unusual to find several clusters for the same region of soil, for the same apparent type of vegetation and so on for other cover types in a scene. These are not only as a result of specific differences in types of cover but also result from differences in moisture content, soil types underlying vegetation and topographic influences. Consequently, a multispectral space is more likely to appear in which each information class is seen to be composed of several spectral classes.

Spectral reflectance characteristics

Representation of information classes by sets of spectral classes

In quantitative analysis it is the spectral classes that a computer will be asked to work with since they are the natural" groupings or clusters in the data. After quantitative analysis is complete the analyst simply associates all the relevant spectral classes with the one appropriate information class.

Classification involves labelling the pixels as belonging to particular spectral (and thus information) classes using the spectral data available. In the terminology of statistics this is more properly referred to as allocation rather than classification. However, classification, categorization, allocation and labelling are generally used synonomously.

Classification Approaches

There are two broad classes of classification procedure and each finds application in the analysis of remote sensing image data. One is referred to as supervised classification and the other unsupervised classification. These can be used as alternative approaches but are often combined into hybrid methodologies

Unsupervised classification

Unsupervised classification is a means by which pixels in an image are assigned to spectral classes without the user having foreknowledge of the existence or names of those classes. It is performed most often using clustering methods. These procedures can be used to determine the number and location of the spectral classes into which the data falls and to determine the spectral class of each pixel. The analyst then identifies those classes afterwards by associating a sample of pixels in each class with available reference data, which could include maps and information from ground visits.

Supervised classification

Supervised classification consists of three broad steps. First a set of training pixels is selected for each spectral class. This may be done using information from ground surveys, aerial photography, topographic maps or any other source of reference data. The second step is to determine error matrix for each class from the training data. This completes the learning phase. The third step is the classification phase, in which the relative likelihoods for each pixel in the image are computed and the pixel labelled according to the highest likelihood.

Supervised classification

The view of supervised classification adopted here has been based upon an assumption that the classes can be modelled by probability distributions and, as a consequence, are described by the parameters of those distributions. As a result it is also referred to as a parametric supervised method. Other supervised techniques also exist, in which neither distribution models nor parameters are relevant. These are referred to as non-parametric methods.

Hybrid Method

It is the use of both classifications techniques i.e. Unsupervised classification followed by Supervised classification method