6 Steps to a Ready to Use,
spatially resolved spectroscopic Diagnostic System

Create a new project and enter an arbitrary name for it. Done!
The system will save all your spectra as well as settings and trained Recognition Modules in this file. This is done automatically, so you never lose your data.

Get a clear overview of all individual examples of one project. If needed, you can add new example data’s or remove them. In this submenu you can organize your examples according to your individual requirements. Simply create your classes, in which the samples should be distinguished, and assign these classes to your samples. Choose one example that you want to edit.

The software PATTERN EXPERT airspect SRS provides extensive visualization capabilities for a detailed analysis of your examples.
Clear arranged visualizations, such as the display of the overview image and the corresponding spectra, admit an exact analysis of the data. The overview screen is a digital color image, which can be converted into a gray one. In addition, the user can select one or more pixels in the picture and display their corresponding spectra. Conversely, the pixels of the selected spectra are highlighted. It is possible to bring out the average spectrum and the area’s standard deviation for all the selected spectra.
Finally, airspect SRS offers to the user the visualization of different images and results for a direct comparison.

The program menu "Configuration" gives the user the ability to perform the various preprocessing steps which are necessary to make your spectra comparable. You may activate and configure a truncation of border areas, a smoothing operation, a baseline correction, and so on. A preview window gives you a visual impression of your settings' effects. The settings will be applied to all spectra in the same manner.

This function of the program allows you to create evaluation processes by using adaptive pattern adaptive pattern recognition algorithms. It is the core function of airspect SRS.
Thereby, the program learns and classifies the spectra, which were previously assigned to a class and selected for learning.
Again in classifying all spectra are used and assigned to appropriated classes. Here the user can choose between two different classification algorithms – classification and clustering. Due to this submenu of the program, an optimal classifier can be created that copes the classification in each class at best.
After teaching and grading an evaluation method – Recognition Module – with pattern recognition algorithms is available for the analysis of new, unknown spectra for the pre-determined question. In the end, the affiliation probability and thus the quality of the results and spectra are displayed on the screen.

The result of the learning and classification process is shown in the program point “Analysis”. The user can draw a direct comparison between the class picture, in which the identified classes of the individual pixels are shown, and the repeatedly portrayed generated spectra image. In this class image the user can see in detail the class assignment probability by the generation of the brightness of the colors, too.