Fast Random Walker Using Precomputation

Provides Matlab code implementing algorithms from the paper "Fast Random Walker with Priors using Precomputation for Interactive Medical Image Segmentation" in MICCAI 2010.

The software uses the random walker image segmentation algorithm with intensity priors to segment images based on user specified seeds. The segmentation time is greatly reduced by performing offline precomputation of eigenvectors of the image graph's Laplacian matrix.

The software provides a method that allows the user to interactively provide seed pixels, see the resulting segmentation, and update the seed pixels accordingly. By turning precomputation on and off, the speed increase it provides is demonstrated, showing that it allows for much better interactive updating.

A manual describing the software is included, and feedback for the authors is encouraged.