1, Use matlab to design improved lenses that are achromatic. (This is by far the best way to go about it, nothing helps better agains aberrations than using an imaging system that has none of those problems. Unfortunately I don't think this is the solution you're asking for.)
2, (This is uggly - because most rgb-detectors use some kind of r-g-b-g-mosaic and then interpolates/inpaints the different channels in some sort of cunning fashions.) You might get somewhere with a 2-step process, where you try to deconvolve away the point-spread-functions separately for the R, G and B channels separately (in principal this works brilliantly, in practice this doesn't at all do as good as adverticed, spelling-error intentional for those who know they know why). Then you will have to spatially transform two of the channels such that they overlap the third. Thay way you have both "focused" the channels that are slightly out of focus (axial) and shifted them to overlap (transversal).
This looks like a neat image-processing project. The problems, as I see them, is that if you're using an rgb-image from a camera with a standard rgb-mosaic, then the in-painting makes for a tricky deconvolution, and the deconvolution is tricky in the best of circumstances. The spatial transformation should in principle be a rather trivial task, but with that type of operation there're often aliasing-problems.
HTH
