Best fit maximizing one parameter while minimizing other parameter

I have two images reference and other is test. I apply model on the ref image to create test images and find out PSNR and AMBE. The model has one parameter 'a' which I am varing. How to find out best 'a' while maximinzing PSNR and minimizing AMBE? In this code I am using multiplication model.

ref=imread('img1.jpg')
a=[1 2 3 4 5 6 7 8];
for i=1: length(a)
    test=a(i).*ref;
    psnrimg(i)=psnr(ref,test);
    AMBE(i)=abs(mean(ref)-mean(test))
end
f=fit(psnrimg,AMBE,'poly5');
%some code here to select value 'a'