by Dr. Nathan Faggian
Python is a great language for prototyping computer vision algorithms, the availability of libraries such as Numpy and Scipy make for rapid development similar to that of Matlab, R and IDL. At the Bureau of Meteorology (BoM) we are solving the interesting problem of weather field warping. Warping (aka non-linear image registration) is used, for example, to determine what the predicted temperature will be hourly if we only have predictions every three hours. Practically, warps (weather advection fields) are estimated using salient image features or data-driven numerical optimizations and this presentation will demonstrate progress we are making on both fronts.
20th–21st August 2011