Opinions on OpenVIDIA CUDA VisionWorkbench? Basis for a general image processing framework?

I’m working on implementing various image processing and fusion algorithms on the GPU and am trying to find a general framework to use.

I’ve been looking at the OpenVIDIA CUDA Vision Workbench (sometimes called CUDA VisionWorkbench, or CVWB), which you can find on the OpenVIDIA homepage: http://openvidia.sourceforge.net/index.php/OpenVIDIA. I haven’t seen CVWB mentioned anywhere in these forums or other forums such as GPGPU, however my first impression is that the source code is of a very high quality.

CVWB provides a set of functions encapsulated as a DLL called CudaVisDLL. While there’s not an exhaustive list of image processing functions, it has quite a few. More importantly it seems that the author(s) have a good architecture in place for managing memory, seem to do a number of optimizations, at least all that I’m aware of, and provide extensive comments.

However, I am only a beginner, and would like to get input from the forum as to the quality and appropriateness of this library for use as a general framework for handling image processing and fusion. Thanks to any and all opinions!
-Josh