04-11-2008 01:34 PM
I am trying to overlay many
small irregularly shaped images onto a larger image.
ImageToImage would be great
except that the copy needs to be masked.
I would like to avoid using full sized
images and masks (many copies, slow).
Is there a better way than pulling out a section, doing a masked copy, and returning it?
Any suggestions?
Thanks in advance,
BrianIMAQ 8.5, C#
04-11-2008 02:43 PM
04-11-2008 03:01 PM
04-11-2008 03:09 PM
04-11-2008 03:31 PM
My original image has many small regions that I wish to process independently. First, I locate the items, then use regions to create a mask. Processing continues using the small mask on a full sized image. A new (smaller ROI) mask is generated and the results are merged back into the original image via the new mask. This is repeated for each of the small items found.
The concern is that all of the processing is performed on full sized images. It works, but I would hope to gain some speed by working on images closer to the size of the area being processed (~1% of the original image).
ImageToImage would work to assemble the results, but bounding rectangles for the small images overlap. Early copies get partially overwritten by the later items.
Thanks for your interest.
Brian
04-11-2008 04:14 PM
04-11-2008 04:19 PM
04-11-2008 07:32 PM
04-28-2008 12:27 PM