Tangles in Images

On our GitHub site we offer two interactive tutorials for how to apply tangles in image processing. The links below lead to descriptive, read-only, summaries of what these two tutorials explore.

The first tutorial finds tangles in images of single letters. These tangles identify the knobbly bits in a letter without cutting it up into segments. They are displayed here as 'fuzzy point sets', or soft clusters, to make them visible. But the point of this tutorial is to exhibit how tangles can identify such clusters without cutting up the image into sets of pixels. As explained in the introductory Section 1.3 of the Tangles book, this is the key idea of tangles from a clustering perspective.

The second tutorial shows how tangles can be used in image segmentation. Here we do cut up an image into sets of pixels, but these sets themselves need not be in any way sophisticated. It is the tangles of many such rough pixel sets that exhibit meaningful portions of the picture. These portions, then, are not precisely defined sets of pixels but, as in the tutorial on letter shapes, more like soft clusters.

While the first tutorial, on tangles in letters, requires less knowledge about tangles, the second is more advanced. What exactly the tutorial require as background is listed on the pages under the two links, in terms of chapters or sections of the Tangles book that are required reading.