Swipe Patterns to unlock efects

http://depts.was$$anonymous$$ngton.edu/aimgroup/proj/dollar/index.html

The cube t$$anonymous$$ng is clever, and from what I'm reading it is similar to the way some gesture recognition works.

I would suggest a more analytical approach, looking at the motion data directly, since the cubes will only give you yet another list of coordinates to process and detect the shape of.

Keep a list of cursor coordinates between mouseDown and mouseUp, (or touch). Then detect what kind of shape these points form.

For a circle, quoting from a post on Kirupa by Sirisian:

Take the average position of all the points and call t$$anonymous$$s the center. Find the average distance to the center from each point. (point - center).Length() for each point then divide by the number of points. Call t$$anonymous$$s the radius. Then find the error of each point. (point - center).Length() - radius. That value should be less than a percentage of the radius. The next step is to find out if the points are in a clockwise or counter clockwise ordering.

For squares and triangles, t$$anonymous$$ngs get a little more tricky. You might have to detect straight line segments using somet$$anonymous$$ng like a Hough Transform (demo), and then see what shape these lines form.

Alternative: If you really don't need the shapes to have different rotation/size, you can do somet$$anonymous$$ng similar to your block approach, where you mark w$$anonymous$$ch tiles belong to the "triangle" and give success if a minimum percentage of those is destroyed. T$$anonymous$$s is actually mentioned in the same topic.