I'm working on a physics lab project where I want to measure the projectile motion of an object. I've asked my students to use the perspective filter to ensure their data is accurate, even if they take the measurements slightly at an angle. However, I'm having trouble with the filter. I've read the documentation, but it's quite brief and left me a bit confused.
In the video we're using, we have a 2-meter stick that we use to measure the size of a region in the image. I want to use the end points of the 2-meter stick to perform the perspective shift. I can align the perspective box in the input rectangle, and when I switch to the output, it appears that the perspective shift has changed.
However, the documentation mentions something about changing the output box to match the scales. I'm not sure what this means or what I'm supposed to do. Shouldn't providing the four points of a rectangle in the image be enough to correct the shift? I'd really appreciate any advice you can offer, as I'm at a loss!
The perspective transform turns a trapezoid into a rectangle which is equivalent to viewing the image face on rather than at an angle, but the ratio of the output rectangle's pixel height to width is not uniquely determined by the transform. If there is an object with known dimensions (square is ideal) then you can adjust the pixel width or height of the output rectangle to match that of the object. Then you can be sure that distances measured along any direction will be accurate. If you only care about vertical distances this last step may not be necessary.
Ahhhh I think I understand now! Sorry, I was interpreting it was the output square was needed for the correction itself, not to give the appropriate units! So if I normally use the coordinate point pair to do my scale, if I use perspective I can just use the output square?
Yes I think from what you describe that you don't need to adjust the output rectangle. But be sure to do the calibration AFTER you have applied the perspective filter.
> So if I > normally use the coordinate point pair to do my scale, > if I use perspective I can just use the output square?
