Based on a grayscale image and an ordered closed polygon (may be concave), I want to get all grayscale values that lie inside a region of interest polygon (as likewise described in SciPy Create 2D Polygon Mask). What is the most performant realisation of that in Qt 4.8? Endpoint should be some kind of QList<double>. Thanks for your advices.
In addition, is it possible to compute a floating point mask (e.g. 0 for outside the polygon, 0.3 for 30% of the pixel area is within the polygon, 1 for completely inside the polygon)? However, that’s just an extra, endpoint would be QPair<double, double> (percentage, value) then. Thanks.
First you need to scanline convert the polygon into horizontal lines — this is done by
getScanLines(). The scanlines are used to sample the image to get all the values within the endpoints of each line, usingscanlineScanner().Below is a complete, standalone and compileable example I had laying around to show that the scanline algorithm is well behaved. It could be tweaked to calculate the fixed point mask as well. So far a point is included if the scanline covers more than half of it in the horizontal extent (due to
round()sinscanlineScanner).Upon startup, resize the window and click around to define consecutive points in the polygon. The polygon you see is rendered solely using the scanlines. For comparison, you can enable the outline of the polygon.
I’m sure the scanline converter could be further optimized. I’m not doing any image sampling, but the
scanlineScanneris there to show that it’s a trivial thing to do.