To calculate relative fluorescence intensity from fluorescence intensity, you simply need some value relative to which you want to have your fluorescence intensity, then divide the other intensities by this value. This value you divide by could be anything: the highest intensity pixel in the image or image stack, the highest possible pixel intensity (255 for 8-bit images), the mean fluorescence intensity in your negative or positive control cells / images, the peak fluorescence intensity in the control image, the intensity of the sixth pixel from left on the second row, or whatever intensity you wish. The sense in calculating RFI depends much on what is it you present your values relative to. You have the same control sample, e.g. well-standardized fluorescent beads, in your every experiment, then presenting the values relative to their intensity could serve to remove some differences that might be hard to control. Even presenting the values relative to the image highest possible intensity could be useful, e.g. if many people have collected a lot of very low resolution fluorescence images using cameras with different bit depths, e.g. 16-bit, 12-bit and 8-bit cameras, and published results using RFI values calculated this way, then you could see how well your data match.
In any case, before you present your image pixel intensity as RFI, you should have an idea why you are doing it that way.