Image hatching

This is one of the end-of-course projects of the Image Processing course taught by Luiz Velho, at IMPA. It was inspired by a paper written by [...: include bibliography].

While the paper worked mainly with real-time hatching of 3D models, we worked with 2D images. Thus, the data about depth and direction is lost, and has to be recovered.

The most important data that can be retrieved from the image are the gradient of the image (the direction that, followed, yields the highest increase on image intensity on a given point) and edges (discontinuities of the image), that delimit, on a real image, the regions occupied by each object.

Sample images

(Each thumbnail links to the enlarged version)

Download

hatching.tgz -- Full program, Python source