Configuring the Preprocessing Settings

Engine Options Particle Classifier Options

Use the settings on the Preprocessing tab to determine how you want the classification engine to calculate the threshold value it uses to segment particles in each image.

Complete the following steps to configure the preprocessing settings:

1. Select a Method for thresholding.
   • Manual Threshold is a method you use to manually determine the threshold range for the ROI.
   • Clustering is a method in which the classification engine sorts the histogram of the image within a discrete number of classes corresponding to the number of phases perceived in an image. Clustering is the most frequently used automatic thresholding method.
   • Entropy is a method in which the classification engine detects samples that are present in minuscule proportions on the image.
   • Metric is a method in which the classification engine calculates a value for each threshold that is determined by the surfaces representing the initial gray scale.
   • Moments is a method of thresholding used for images that have poor contrast.
   • Inter Variance is a method of thresholding used for images in which classes are not too disproportionate. For satisfactory results, the smallest class must be at least 5% of the largest one.
2. If you selected Manual Threshold in Method, use the Min and Max controls or the slider at the bottom of the histogram to set the threshold value. If you select an automatic threshold method, the algorithm calculates the threshold value and updates the Threshold Range. You can specify a Lower Limit and an Upper Limit for the threshold computed by the automatic method.
3. Select a type of object to search for from the Look For drop-down menu. The NI Classifier will look for Bright Objects, Dark Objects, or Gray Objects.
4. Click Reject Objects Touching ROI to ignore objects that are touching the border of the ROI you drew.
5. In Remove Small Objects (# of Erosions), select the number of erosions you want the classification engine to perform to remove small objects from the ROI.

   As you manipulate settings on the Preprocessing tab, the NI Classification Training Interface displays thresholded objects in blue.

Tip  If the samples in the ROI are not classified as you expect them to be, experiment with settings on the Engine Options and Particle Classifier Options tabs to improve the classification.