If you don't want to used computer based software, you can do it field by filed by estimating the % of stained cells X by the color intensity. However, this method is know to be subjective and thus it is preferred to ask 2 blind reviewers to do it to try to decrease intra- and inter-observer variations. If they disagree, you need to consult a third examiner to evaluate and report the percentage and amount of staining.
ImageJ/Fiji offers some great plugins that might be of use if you know the histological stain. One of them is called "Colour Deconvolution" and should pretty much be what you are looking for - see details on the link provided.
Quantification of a specific colour within the stain can be done with thethresholding following the deconvolution as well.