So i've been thinking of a way to detect various fonts, and i have an idea. Instead of automatically creating a graph of specified cell size, i should scan for light colored columns of pixels between letters to determine their width. Once i have determined their width, i can then scan them vertically within that pixel width to determine the height of the letter. This also adds the extra benefit of using fonts that don't have a specific width set for each letter (verdana, times new roman, etc.).
Do you guys see any problem with this? Any other ideas? Let me know.