Rob, I checked out this post as follows...
What you are doing here is generating the low pass version and removing that from the image leaving the high pass only which would even out the brightness. Its that step of desaturating thats a bit odd. What I reckon you need for this technique is to take an image and call it A. Clone it to B and again to C. Take C and blur it so massively that its one uniform shade. Take B and remove C from it - thats sort of like the ultimate desaturation. Take that result and blur it a little and then remove this image from A. Then you would be removing the low frequencies but not the DC component so it should give the high pass but keep the DC colour. Its the same as doing the high pass and adding back in the DC of course as its A - (B - C) which is A - B + C. I'm still not sure how you can actually achieve this without the colour values going negative which would be clamped and then it wont work correctly. The only way I can see of getting around this is to split the images up into parts lighter than DC and parts darker than DC and process separately. I'll try this out offline and see if I can get that to work.