Landmark Correspondences

The plugin Landmark Correspondences calculates a transformation between two corresponding landmark clouds and renders a transformed image. The landmarks are read from point selections over two images. The transformation is estimated by a least squares or Moving Least Squares fit for the available models.

The non-linear non-invertible transformations as estimated using the Moving Least Squares method are rendered through a mesh of triangles whose resolution is a parameter of the plugin, see Interactive Moving Least Squares for an intuitive explanation.

Two snapshots from the ImageJ Conference 2008 registered into each other using an affine transformation as estimated from automatically extracted Feature Correspondences.

A photograph and a cartoon registered into each other using the Moving Least Squares method and a similarity transformation as estimated from manually set landmark correspondences.