Autostereogram has been a fantastic media since it's debut. I will present
an algorithms for efficient reconstruction of binocular disparity map from
autostereogram and interaction with autostereogram. Accurate
reconstruction of binocular disparity map from autostereogram is an
ill-posed problem. We formulate the reconstruction problem as an energy
minimization process which is appropriate for utilizing graph cuts
technique. With the observation about binocular disparity in
autostereogram and piecewise smoothness assumption of reconstructed
surface, we achieve effective recovering of disparity map with several
cycles of graph cuts. With our robust and fast reconstruction algorithm,
it becomes possible to interact with autostereogram, which is a novel
feature in understanding autostereogram.