How to fill in a region with different patterns?
in [Matlab]
|
From: Zuofeng on 21 Feb 2010 15:31 Hi to all, Here is a question which I cannot solve. Appreciate so much for any suggestions! I have a squared region which is irregularly divided into many rectangular patches. Each patch is associated with a value, and two patches possibly share a common value. I hope to fill in each patch a pattern according to its value. For instance, if a patch has value 1, then I fill in that patch with pattern A; if it has value 2, then fill in it with pattern B,... The pattern is something like a rectangular region with certain colored lines in it. For instance, pattern A may be a white rectangular region filled in by two thick black lines; pattern B may be a white rectagular region filled in by 10 thick black lines,... Is this available in matlab? I have serached "polygnon" in matlab user guide and found a "fill function" may work, however, there is still a long way to finally complete it. Many thanks again! Zuofeng
From: ImageAnalyst on 21 Feb 2010 15:56 Zuofeng Here's what I would do. Make each pattern the size of the entire image. Then for each patch's gray level, threshold the image at that gray level and use logical indexing to assign the output image the values from the pattern image that you want to use. Make sense? -ImageAnalyst
From: ImageAnalyst on 21 Feb 2010 15:58 Post your input image, and your pattern images to http://drop.io.
From: Zuofeng on 21 Feb 2010 21:21 Hi ImageAnalyst, Thanks a lot for response! This idea maybe works, but I still have two questions. 1. Suppose I have made pettern A an image the same size as my original image, then how do I threshold it at the gray level of A? My concern is that it is possible to threshold only a part of pattern A which is not my original pattern A. 2. How do I assign the output image a logical index? There may be many values from a pattern, how do I assign them all to an output image? Please forgive me if I have misunderstood your idea! Many thanks! Zuofeng ImageAnalyst <imageanalyst(a)mailinator.com> wrote in message <be2cf1bd-43e1-4d28-b452-5026aaac1385(a)b10g2000vbh.googlegroups.com>... > Zuofeng > Here's what I would do. Make each pattern the size of the entire > image. Then for each patch's gray level, threshold the image at that > gray level and use logical indexing to assign the output image the > values from the pattern image that you want to use. Make sense? > -ImageAnalyst
From: ImageAnalyst on 21 Feb 2010 22:46
1. You have to decide where the "patches" or regions are. Thresholding is one way, such as this way to get a binary image of where all the pixels with value 4 are: % Generate some sample data in the range 0-9. a= uint8(10 * rand(8,8)) % Get a logical array of pixels with a value of 4. pixelsOfValue4 = (a == 4); Or you could pick a range like pixelsOfValue4through6 = (a >= 4) & (a <= 6) ; 2. This demo should explain it: % Generate some sample data in the range 0-9. a= uint8(10 * rand(8,8)) % Get a logical array of pixels with a value of 4. pixelsOfValue4 = (a == 4); % Reassign to 99 instead of 4. a(pixelsOfValue4) = 99 |