Hi,
is there a way to load an image file (.jpg or .bmp), "scan" it, with a defined velocity, so to convert a picture in numbers (practically to take the coordinates in number boxes)?

The image could have a black background (ignored by scan process), and lines or points drawed in white (but also in other colours..).
Then I would use the numbers taken from those white sketches (lines, points, rectangles, etc.) as frequencies, so that a curved line is converted in a "glissando", a point in a "pluck", a rectangle in a "cluster", etc.

Is there a patch about this?
Any suggestions?

Thank's,
t.

It's easy enough to transform a bitmap image into numbers outside of pd (using c++, java, c#,...) and save those numbers so you can read them into a pd table. Recognizing rectangles and lines might be a bit harder. If you want to do it all in pd i assume you should be looking at gem.

gridflow might be a good choice too, it has a photo pianoroll example, no need for the opengl complication