Reducing Computation Time in a Reaction-Diffusion Stereo Algorithm

A reaction-diffusion stereo algorithm consists of multi-sets of reaction-diffusion equations. Since the equations are described with time-evolving partialdifferential equations, they require much computation time. In addition, the previous algorithm does not state any criterion for convergence judgement; we need to compute the equations for enough duration of time until their solutions converge. In this work, for reducing computation time in the algorithm, we propose a criterion for convergence judgement and implement the algorithm on a multiprocessor computer system.

[1]  Hidetoshi Miike,et al.  Reaction–diffusion algorithm for stereo disparity detection , 2009, Machine Vision and Applications.

[2]  Pavel Mrázek,et al.  Selection of Optimal Stopping Time for Nonlinear Diffusion Filtering , 2001, International Journal of Computer Vision.