Classifier

This module implements methods to detect patterns in image regions. Commonly, classification algorithms are divided by the level of classification (pixel or region), and by the interaction of the user (supervised or unsupervised). Pixel-based algorithms classify individual pixels according to their resemblance to a specific pattern.  Region-based algorithms use regions from segmented images, and classify each region to a specific pattern. Supervised methods uses a predefined typology, given by the user, who supplies samples of each pattern. Unsupervised methods detect an unknown number of patterns, according to their own method.

The available methods in TerraLib are:

• ISOSeg,

• Expectation-Maximization - EM,

• K-Means,

• Spectral Angle Mapper - SAM,

• Maximum a Posteriori Probability - MAP.

ISOSeg

This is an unsupervised and region-based classification algorithm.

Input:

• Raster

• Vector of polygons

• Acceptance threshold

Expectation-Maximization - EM

This is an unsupervised and pixel-based classification algorithm. Expectation-Maximization (EM) works iteratively by applying two steps: the E-step (Expectation) and the M-step (Maximization). 

Input:

• Raster

• The value of “K”, which stands for the number of patterns (or clusters) to find in the image.

• The maximum number of iterations (E/M steps) to perform if convergence is not achieved.

• The maximum number of points used to estimate the clusters (default = 1000).

• A convergence threshold. When the clusters change in a value smaller then epsilon, the convergence is achieved.

• The previously estimated means of the clusters (optional).

K-Means

This is an unsupervised and pixel-based classification algorithm.

Input:

• Raster

• The value of “K”, which stands for the number of patterns to find in the image.

• A convergence threshold. When the clusters move less than this threshold, the algorithm stops.

• Maximum number of iterations.

Spectral Angle Mapper - SAM

This is an supervised classification algorithm.

Input:

• Raster

• A set of ROI samples.

Maximum a Posteriori Probability - MAP

This is an supervised classification algorithm.

Input:

• Raster

• A set of ROI samples.

Maximum Likelihood/ICM (a.k.a MaxVer/ICM)

This is a supervised classification algorithm.

Input:

• Raster

• A set of ROI samples.

Parameters:

• Apply ICM: Enable/disable the use of Iterated Conditional Modes (ICM) after the maximum likelihood estimation.

• Maximum of iterations: Maximum number of ICM iterations to perform.

• Iterations Threshold: Minimum number of changed pixels (percentage) between iterations. If the number of changed pixels is below this value the process is stopped.

References:

• On the Statistical Analysis of Dirty Pictures - Julian Besag - Journal of the Royal -Statistical Society. Series B (Methodological), Vol. 48, No. 3. (1986), pp. 259-302.

• ERTHAL, G.J.; FRERY, A.C. Segmentação de imagens multiespectrais pelo algoritmo Icm: Integração ao ambiente spring. In: SIMPÓSIO BRASILEIRO DE COMPUTAÇÃO GRÁFICA E PROCESSAMENTO DE IMAGENS, 6., 1993, Recife Anais...Recife. SIBRAPI, 1993. p.33-36.

It is accessible through:

   Raster Processing > Classification... (list of all raster layers will be available) 

This wizard consists of the following steps:

    Wizard Page 1 - Selection of the layer to execute the operation (Layer Search)

1. On the List of Layers select the raster layer to apply the operation. 

2. Optionally use Filter By Name field giving part of the layer name to help find the layer in the list.

3. Press Next  to go to next step or Cancel to close the dialog. 

    Wizard Page 2 - Classification parameters

1. Select the type of classifier to be used.

2. Select the bands to be used in the process.

3. As described above, each classifier has a set of specific attributes. For supervised classifiers (SAM and MAP)  is necessary to use a component for the acquisition of samples.

   • Use  tool to enable the acquisition of components (samples) through the interface Raster Navigator. This interface offers several tools (such as zoom, color composition) to help on collecting good samples. 

   • Use    tool to create a region of interest over the image that represents a homogeneous region of a desired sample.

4. Press Next  to go to next step, Back to return to the previous wizard or Cancel to close the dialog.

    Wizard Page 3 - Output information

1. Raster Info - First press and inform the folder where the resulting file will be saved.

2. Name - inform the raster name.

3. Extra Parameters - if there are some, see the details on how to inform then here.

4. Press Finish to save the resulting contrasted raster or Back to go to the previous wizard page.

Hint: The resulting image will be added as a new layer at the TerraView project.