Pixel art is a particular form of digital art where pixels are used to represent the different features in an image. Almost all arcade game graphics were done using pixel art. At the time they were created, image size had to be quite small because of memory and space limitations. Nowadays, there is no such limitation; however, converting a pixelized image into a resolution-independent vector representation is a complex task.
The authors of this paper propose a novel technique for performing a conversion from pixelized image to vector image that produces impressive results. “Ambiguities in the connectedness and separation of diagonal neighboring” pixels that are only connected through a single pixel are resolved by converting them into regions with smoothly varying shading and separating them by piecewise smooth contour curves.
The paper is very well written, and the organization makes it easy to follow. It should be an interesting read for those interested in computer graphics, as well as the wider graphic arts community.