[Dclib-devel] Learning a kernel for image transformation

[Mayuresh <may...@ac...>]

Hi.

I am new to dlib and ML.

I wish to solve a problem where a grayscale image (a digital X ray to be
precise) is enhanced by a vendor software to improve its sharpness.

I am not able to achieve similar results using interactive tools like
gimp.

So I wish to machine learn that image transformation so that I can apply
it automatically to other raw images to get equivalent results.

Is it possible to learn a kernel to do this and any advice on how to go
about it?

I see a lot of examples for image classification, but none to learn an
image transformation.

A sample raw image: http://mayuresh.sdfeu.org/bypyrvg.png
Processed by a proprietary software: http://mayuresh.sdfeu.org/byvendor.jpg

-- 
Mayuresh