Geek asks and answers

Template compatibility - different template and image sizes

I have an image that I can get with better quality (for example, a higher size). The problem is that I cannot perform pattern matching by default size, only at a higher level. And the top one is not fixed - it can be anything from 1.0 to 2.5x from the original image.

I thought about extracting templates in higher resolution, and then, depending on the size of the image, resize the templates. However, this can lead to completely incorrect results when I use a different resizing algorithm.

What are you offering? How should I continue when I want to perform pattern matching in such circumstances?

Edit: just mark - the templates will always be the same, and their copies will also be displayed on the images (images are generated by the computer).

+3
source share
1 answer

So, you say that you have an image template in a nominal size of 1.0 and a target image in which you want to match from 1.0 to 2.5x?

Algorithms for matching invariants on a scale scale are rather complicated. The simplest thing is to scale your target image down to a few intermediate resolutions and try to match it with your template. For best performance, try downscaling your template (say, 0.5) so that your target falls in the range of 0.5-1.75. When working with different resolutions, try to smooth out the images a bit.

, , , , {x, y} res. , . - .

+1

More articles:


All Articles