Are you sure you've got that formula right? That's a very difficult integral to do (I'm not even sure it's possible to do without a computer).
Edit: to the poster above me, It's possible to do using wolfram: manipulate the integral then use wolfram to integrate cos (x^2), which gives you a result in terms of the Fresnel Integral, or just put numbers in instead of a and b and wolfram stops being confused e.g. http://www.wolframalpha.com/input/?i...8x-3%29%5E2%29
After trying a few things I can't see a way - the and are tricky to sort out if they're not equal (though someone else may be able to find a way, there might be a contour integral which can sort it out)
Might be best to try some numerical methods if you have the values of the constants?