Derivatives

Concerning feature wishes, I would very much like to experiment with a derivative (or image gradient, in the technical sense) component. Ideally, it should be able to return its results in X-Y-component form as well as in norm-orientation-form. Some Gaussian prefiltering as performed by a Sobel kernel would be nice.

Moreover, nonlinear filters such as median or dilation/erosion would be cool. I see, however, that those are intricate to intergrate in a resolution independent framework. However, the Gaussian component in Filter Forge has some kind of kernel size parameter as well, hasn't it?

Well, and the coolest of all would be a C++ component API, ideally with an image format compatible to the Intel Image Processing Library used in OpenCV. This could make FilterForge the numer one prototyping tool for algorithms development in image processing!

David