Training the Classifier

Complete the following steps to train a classifier file by manually classifying a sample image into a new or existing class.

1. Select File»Open Images.
2. Navigate to the image that you want to use as a sample, then click OK.

   Tip  You can select all of the files in a folder by enabling the Select All Files option, or you can select multiple files by holding the <Ctrl> key while clicking file names. When you select an image from the list, the Preview Image window displays the image, file type, image size, and image type. If you select a collection of images, the Preview window displays all images in a sequence. To view the sequence at a different rate, adjust the slide to the right of the Preview Image window.

3. If you open multiple images, use the navigation buttons to browse to a specific image.

Add Samples Tab

Use the Add Samples tab to apply a class label to an image or portion of an image. The image is then added to the classifier as a sample of the applied class.

1. If the image contains multiple objects, draw an ROI around the object you want to add as a sample.
2. Adjust the parameters on the Preprocessing, Engine Options, and Particle Classifier Options tabs, as necessary.
3. Assign a class label to the sample using one of the following methods:
   • To assign a new class label, select Add New Label from the Class Label list. The Enter a new label dialog box opens. Type the label name and click OK. The new class is assigned to the sample.
   • To assign an existing class label, select the label from the Class Label list. Click Add Sample. The class is assigned to the sample.

Classify Tab

Click Train Classifier to train the classifier engine.

Note  If you add new samples, you must retrain the classifier.

Note  The Class Population histogram displays existing classes and the number of samples for each class. The Total # of Samples indicator displays the number of samples you have identified for all classes.