PatMax

Unlike the other CVL vision tools that support nonlinear client coordinates, PatMax is a feature-based rather than pixel-based tool. PatMax works by locating a pattern of features in the run-time image that matches a trained pattern of features, then returning the transformation that describes the relationship between the trained pattern and the pattern in the run-time image.

If you supply a run-time image that has a nonlinear client coordinate transform, PatMax computes the best-fit linearization of the client coordinate transform across the entire input image, then applies that transformation to the features in the input image before locating the trained pattern.

Note: Nonlinear client coordinates are fully supported for a pattern training image as well. If you supply a pattern training image that has a nonlinear client coordinate transform, PatMax computes the best-fit linearization of the transform across the entire input image, then applies that transformation to the features in the input image before training the pattern.

After PatMax locates the pattern instance or instances in the run-time image, it computes the best-fit linearization of the nonlinear client coordinate transform at the center of each found pattern instance (where the center is the center of the smallest rectangle that encloses all of the pattern’s features), and it maps the returned pose through that local linearization before returning it.