How to convert individual x y maps to 2-channel matrix map in OpenCV?

Question

I am trying to do forward mapping with OpenCV, but remap() only allows inverse mapping so I need to inverse the map(s) before using them. This is the inverse map method I want to use and it only takes in a single 2 channel map.

In the convert maps docs it says that you can convert one 2 channel (x,y) map into individual x and y maps. It also says that you can do "Reverse conversion". Does this mean you can convert from x and y maps to a single (x,y) map shape?

Christoph Rackwitz · Accepted Answer · 2022-06-15T07:58:28.170

0

import cv2 as cv

(xymap, _) = cv.convertMaps(map1=xmap, map2=ymap, dstmap1type=cv.CV_32FC2)

Documentation isn't terribly obvious. The bullet list only contains the interesting cases (for fixed point math). Your case is just rearranging the values without touching them, which is not listed, but supported. You can use dstmap1type=cv.CV_32FC2.

Or simply use numpy:

xymap = np.dstack([xmap, ymap])

You are working with v2.x documentation... don't. Use current docs. You are very surely not using OpenCV v2.x.

edited Jun 15 '22 at 07:58

answered Jun 15 '22 at 01:05

Christoph Rackwitz

11,317
4
27
36

could you modify the inverse map method I linked so that it can take just a 1d map instead of xy? – monopoly_lover Jun 15 '22 at 01:19
but you can simply use a combined x-y-map? why would you need that code modified? if you wanted to modify that code, where are you stuck? – Christoph Rackwitz Jun 15 '22 at 01:22
@ChristophRackwits Nevermind your right I dont need it. Thanks for the help! – monopoly_lover Jun 15 '22 at 03:03
I read the docs again. you're absolutely right, they don't actually list your use case, even though the given list suggests totality. it is implemented though. – Christoph Rackwitz Jun 15 '22 at 08:04

How to convert individual x y maps to 2-channel matrix map in OpenCV?

1 Answers1