Through the looking glass

One has, after much agonizing, decided to reveal one’s visage on this blog. It seemed like a pity to hide a jaw so finely chiselled, cheekbones so gloriously high, eyes so piercingly erudite, dimples so burstingly deep behind this hypertext curtain of blind, impersonal anonymity.

Behold.

Wotsay, huh? Stud or what?

This is a picture of oneself. God promise. Except it’s been transformed. Instead of the x- and y-coordinates representing vertical and horizontal distance, they represent the bearing and distance of a line from an arbitrarily-chosen origin. Loosely speaking, each point in the transformed image corresponds to a potential line in the original, each sinusoid in the transformed image corresponds to all the lines passing through a point, and places where many sinusoids intersect are distinct lines in the original. Or something like that.

This is the Hough transform applied to the problem of line-finding. The same transform finds application in the reconstruction of slices of the human body from X-ray projections. In case you’ve ever wondered (er .. if you haven’t, you may proceed to wonder now) how they see slices of your insides during a CT scan.

And a transformed image invariably looks pretty good too, in a snapshot-of-the-very-fabric-of-reality kind of way.

Yeah, anyway, so the bright spots represent lines in the original image. There are six of them here (yes, you count eight, but the left of the image is supposed to “wrap round” to the right) so apparently one has a hexagonal face. Funny, actually .. one always thought one was rather square. Maybe it’s the dimples.

*adjusts geek-glasses and shuffles off*

[Note: Transform implemented using image processing toolbox in MATLAB™. Thanks to The MathWorks for developing a useful tool.]