Reverse Digital Typography
In this project, we study the problem of reconstructing parametrized fonts from scanned images. In the first part, we investigate high-resolution noise-free images. In the second part, we study the noisy images, where artefacts such as low-resolution and blurry edges come into play. We present the quad-tree spline fitting method to robustly reconstruct parametrized fonts in the presence of print-scan artefacts.