|
|
Typeface recognition |
Font Moose attempts to recognize the fonts used in images of letters or words. Launched in January 2015, the on-line software will be in beta mode for some time. Initially, the results are limited to the fonts available at MyFonts. For fonts not in that data base, Font Moose tries to give the closest approximation. |
![]() |
|
Who are we? |
Luc Devroye and Harry Lowson, School of Computer Science, McGill University, Montreal, Canada. ⦿ lucdevroye@gmail.com ⦿ harrylowson15@gmail.com ⦿ http://luc.devroye.org/index.html ⦿ http://luc.devroye.org/fonts.html. |
Better programs |
Font Moose is a miserable piece of software. For professional help,
please try these fine services:
|
Apologies |
Our data base uses public pixel images of limited accuracy. We can't perform miracles with those. |
The software |
We offer three ranking and classification algorithms:
|
Ranking |
Font Moose is based on an in-house algorithm that uses methods from computational geometry, Fourier analysis, probability theory and machine learning. The ranking of letters is driven by the cacotropic index, a number between 0 and 1 specially developed for this application. For words, an amalgated index is computed whose value may occasionally exceed 1. |
Contact |
Luc Devroye |
![]() |