perClass Mira provides numerous ways how to extract and export information from a single or multiple images. The use case is to define classifier, segment objects or specify regions of interest and export user-defined features to external file (Excel .xls or XML). This data is the used for custom data analysis or further research.


Let us walk through a basic example using Feature extraction panel. In order to extract data, we need to specify

  • Where the data is extracted from
  • What pixels are included in the extraction
  • What is being extracted



In the first example, we want to extract mean spectra from objects. Therefore, we select Mean spectrum from the Add representation combo box.



You may select multiple representations  


Select one of more images in Images list and then File  menu / Export and Export region features to Excel. Note you may also export the same data into XML. That option is more convenient if you wish to programmatically post-process data analysis.



You will asked for a name of a file to save. By default, perClass Mira exports to XLSX format. This allows for more than 256 columns which is useful if we're exporting a lot of features per object, for example mean spectra. If you prefer the legacy XLS format, you may choose it in the export dialog.


In the screenshot below we can see the structure of the exported data:



Objects of each selected image are described by rows. For each object, we can see the scan name followed by an object name. When exporting object segmentation, the object name is automatically assigned in the segmentation procedure. When exporting content of user-defined regions, the region name is used. This can be user-assigned. For each object, its bounding box and per-object decision is provided.  The Content present column shows whether there is content represented in this object/region and if so, how many pixels. When exporting objects, the content is always present. When exporting regions, this may not be the case.

Finally, the section contains the exported data. In our case, the columne correspond to individual wavelengths of the mean spectra extracted for each object.