Using a Specified Bright Field Image for Flat Field Correction

Select Correction with specified bright field image to use one image as the bright field for flat field correction. Click to launch the Flat Field Creation Wizard to create a new bright field image.

Use this option if you can acquire an image of only the background.

You can use a single state in Vision Builder AI to correct new images from the acquisition step or the flat field step, as shown in the following diagram:

Using Online Learning for Flat Field Correction

Select Online Correction Using Modeling for the inspection to relearn the bright field at each iteration. Use the Estimate Background controls to configure how the algorithm computes the bright field.

This option is recommended when background variations are inconsistent across images.

Using Online Learning with a Simple Flat Field

You can use branching or a different inspection to acquire and/or process a background image, then use the newly computed flat field image for future corrections. This setup creates a background image to compute the flat field image, then saves the flat field image to correct the image from the next iteration.

In this model, online learning is not done at each iteration. The following diagram shows online learning with a simple flat field:

Using Iterative Online Learning with Modeling

You can create an inspection that acquires an image, learns the flat field model from that image, then corrects and processes the image in one state, as shown in the following diagram:

This setup does not save the flat field image.

Using Online Learning with Modeling

You can create an inspection where the flat field is not learned at each iteration, but is applied to the acquired image, as shown in the following diagram:

Using Different Inspections to Process and Save Flat Field Images

You can create two separate inspections to process flat field images, as shown in the following diagram: