Yes, you can certainly use getPerspectiveTransform and warpPerspective functions if you know where the wanted corners are in the input image (like the corners of the credit card in your example.) It will work using either C or Python API. It all boils down to the question on how you can find the four corner points for the transform.

Cheers,
-- Keijo

Hi Lukas,

In order to do perspective correction, you first have to calibrate the camera. In general it is not possible to just take a picture with any camera and change the perspective into something else without any knowledge about the position of the camera with respect to the object of interest.
One option is to use a QR code for the calibration. The method is described in this Visionappster cookbook. It would obviously be a bit challenging to glue a giant QR code on the wall of a 32-story building like in your example image so you'd have to tone down the scale of your scene a bit

Cheers,
-- Keijo

Hi,
It looks like you have done everything right. Just click on the white Play button at the bottom left corner and the image will appear at the output.

Happy experimenting
Cheers,
-- Keijo

@luqt123
Hi Lukas,

Assuming that you have already allocated space for your kuvio_image, you can copy it across using OpenCv functions. Possible functions for this are for example cv::cvtColor, cv::mixChannels and cv::merge. Please read OpenCv documentation on how they are used.

For example, if your opencv image is a 24-bit RGB image, the copying can be done simply as follows:

int my_opencv_function(/*params,*/  kuvio_image* imgOut)
{
  if (imgOut->type == kuvio_rgb32_image_type)
    {
      cv::Mat myOpenCvImage = ... // Get or create an RGB image. eg. cv::imread
      /// ...do stuff with the image and copy to kuvio_image
      cv::mixChannels(myOpenCvImage, OpenCvMat(imgOut, OpenCvMat::Rgb24), {0, 0, 1, 1, 2, 2});
}

Please bear in mind that the sizes of the allocated kuvio_image and the size of the image in cv::Mat myOpenCvImage must be identical.