Statistics of the Samples

After executing the operation Image Statistics... on the main menu and obtained the desired sample, two windows will be presented: one with the numeric statistics calculations and another with a graphic visualization.

For statistic parameters such as Moments, Median and Covariance and Correlation Matrix, the graphic presented corresponds to a histogram of the selected Infolayer. For Autocorrelation and Cross-Correlation Matrixes, the graphic will show the corresponding curves of these functions related to the selected Infolayer.

The fields Operation and Samples show the statistical parameter and sample name previously selected by the user.

Depending on the selected Calculation, you will obtain the following outcomes:

Moments

The Statistics of Samples window will present the following statistics for the selected bands: mean, standard deviation, variance, moment 3, moment 4, mode, minimum value, maximum value, Variation coefficient, asymmetry coefficient and kurtosis.

Median

The Statistics of Samples window will present the median value for the selected bands.

Covariance and Correlation Matrix

The Statistics of Samples window will present the covariance and correlation matrixes for the selected bands .

These values are shown in the same order as the bands are in the Layer Selection list. For example, if the images are presented on the selected list in the following order:

band 3

band 4

band 5 ,

the Covariance and Correlation matrixes will be presented as follows:

33 34 35

43 44 45

53 54 55

where ij correspond to covariance or correlation values between bands i and j.

Correlation Matrix

The sample autocorrelation values are calculated according to the point neighborhood for different distances measured in number of pixels. These distances are defined by the LAG (4-20) parameter on the Obtaining Samples window. Therefore, the autocorrelation values are calculated for all distance values on the range [0,LAG]. The outcome is presented as a matrix and a graphic, showing the curves of the autocorrelation function only for X and Y.

Cross-Correlation Matrix

When you choose Cross-Correlation Matrix, on the Obtaining Samples window, a new Infolayers window is presented and you should:

  • select two or more bands (ILs);
  • click on Apply .

Similarly to the autocorrelation matrix, the cross-correlation values are calculated according to the point neighborhood for different distances measured in number of pixels. The difference is that the neighbor points are caught on different bands, taken two by two. Thus, the correlation values are calculated for all distance values between [0,LAG]. The outcome is presented as a matrix and a graphic, showing the curves of the cross-correlation function only for X and Y.

The outcomes can also be stored on disk for further analysis .

  • Saving the outcome on disk:
    • click on Save , choose the directory you wish and type in a name for the file.

Going back to the Sample Statistical Analysis window, you will be able to:

  • Loading and Updating a file:
    • select a directory and a file on the list Statistical Files;
    • click on Load ;

On the field Name, you will see the filename selected;

The information layers will be updated on the Infolayers list.

  • click on Update .

The dialog box Obtaining Samples will be presented. Its operation is the same as described above.

  • Deleting a file:
    • select a file from the list Statistical Files ;
    • click on Delete .

After creating or loading a statistical file, it is necessary to save it for further use. This file will contain the ILs and obtained samples.

  • Saving a file:
    • click on Save on the Sample Statistics Analysis dialog box.