This document describes the HydrOffice BRESS (2.2). For the project page, go here.

2.3. Segments

This tool segments the seafloor based on elevation and reflectivity values. It extends the functionalities provided by the Landforms tool (see Landforms) by analyzing the reflectivity values co-located with the elevation nodes that belong to each area kernel.

2.3.1. How To Use?

  • Select the Segments tab (Fig. 2.14) on the bottom of the BRESS interface.

  • In Setting, check the settings of your choice.

  • To change the Settings for Segments v2:

    • Click the Unlock button, and click OK to the dialogue.

    • The parameters in common with the Landforms tool are described in Landforms.

    • The reflectivity Histogram (see Reflectivity Histogram) can be customized by modifying:
      • The min and the max fields that represent the minimum and the maximum values for the histogram’s bins.

      • The nr.bins field sets the number of bins in which the range between the minimum and the maximum values on the histogram are split.

    • The range percentage and the delta bins fields are used in the Splitting processing step (see Area Kernels Splitting) to identify peaks in the reflectivity histogram of each area kernel.

    • The intersection and the min.samples fields are used during the Merging processing step (see Area Kernels Merging) to identify the area kernels that can be merged together due to similarity in the reflectivity texture.

    • In addition to the Landforms products, the tool can provide the following three kinds of outputs (and they can be saved as ASCII grids or plotted):
  • To reset the Parameters to the default, initial values, click the Reset button.

  • In Execution, click Segments v2.

segments tab

Fig. 2.14 The Segments tab.




2.3.2. How Does It Work?

2.3.2.1. Reflectivity Histogram

The Reflectivity Histogram is used to characterize the texture in each area kernel.

2.3.2.2. Area Kernels Splitting

The splitting step evaluates the reflectivity histogram of each area kernel to identify multi-modality (that is, multiple peaks) (see Fig. 2.15).

The range percentage value is used to identify a threshold density (as a percentage of the dynamic range of each histogram) for classifying potential peaks (and, thus, multi-modality) in an area kernel.

The delta bins value represents a filter that removes peaks that have an higher peak in the surrounding histogram bins.

splitting

Fig. 2.15 An example of reflectivity histogram with two detected peaks.

2.3.2.3. Area Kernels Merging

The merging step compares pairs of reflectivity histograms to detect similarity in reflectivity textures (see Fig. 2.16).

The intersection value provides the criterion used to evaluate if a pair of reflectivity histograms are similar enough to be classified as part of the same segment.

The min.samples value is used to ignore the area kernels that are too small in size to have a reliable reflectivity histogram.

merging

Fig. 2.16 An example of histogram comparison. The pair of reflectivity histograms has an intersection value of 71.56%.

2.3.2.4. Segments

A segment represents an area that has been classified with the same landform type and similar reflectivity texture.

2.3.2.5. Segments Statistics

The statistics for each segment are optionally calculated and stored in a .csv file.

The statistical values (i.e., median, mean, and standard deviation) are based on the mosaic values and the Statistical layers of all the valid nodes in a segment.

2.3.2.6. Mosaic Mask

The mosaic mask has the same information as segments, but resampled at the resolution of the input reflectivity grid.

It is usually used for theme-based seafloor characterization.

2.3.2.7. Cropped DTM

The cropped DTM is obtained by removing all the unclassified nodes from the input elevation grid.