BryanBentz Posted July 14, 2014 Share Posted July 14, 2014 It seems I'm often working with rectangles, or images of rectangles - e.g., a scanned photo. Sometimes the photo is at a small angle - to date I've been using the Layers/rotate menu item to try to adjust the image is 'square' by hand, and while I'm careful I'm not perfect. It would be great if this could be automated; it would likely require some edge detection to pick out the rectangle, with an exact angle of rotation then being computed. It's possible this might be done with a plugin, but that might not be the best route - it's a sort of global image change, not a pixel-by-pixel local transformation. I did look through the index of plugins (found some neat stuff) but didn't see anything like it. If one wanted to guild the lily one might imagine an image from a flatbed scanner with a number of photos on it; identifying these and rectifying them, perhaps moving them to separate images automatically, would be enormously helpful. Finally, one transform that would be lovely to have is what I nickname the 'billboard' transform - imagine that you take a photo of a billboard on a highway, and it is at some angle. If you could identify the corners of the billboard in the image (and this will be a trapezoid), it'd be nice to then have a transform that 'squares up' that image so it is as if you're looking at the billboard straight on. Some time ago I looked at coding this myself; there's a transform, the Hough transform, that seemed to be the thing to use, but it always seemed underconstrained to me (e.g., it involves one additional scale factor that seemed completely arbitrary). (I shouldn't have said 'finally' above). One other thought - there is a free software tool out there, SCILAB (http://www.scilab.org/scilab/about), that is pretty good for writing image processing algorithms (and there are lots of algorithms written in it). I wonder if it would be possible to establish some sort of integration with Paint.net, or at least an interface (so that, for instance, SCILAB could grab the image from Paint.net, operate on it, and push it back to Paint.net). One can of course just invoke SCILAB on a saved image, but having it integrated into the user interface would let an entire new range of plugins to be accessible interactively. It might be one of those steps that would pay off a great deal. Quote Link to comment Share on other sites More sharing options...
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.