Giacomo's profile - activity

Found the problem. It was exactly as expected :)

After several tests, I believe that I am using the function incorrectly. Probably I am passing some argument in the wrong way.

To be sure to use the function correctly I compared the results of the opencv triangulatePoints function with your function. The results for two points should be quite similar but in my case they are not: opencvTriangulate: [-169.1275598 157.77955505 19.13403159] YourFunction [ 150.95142557 -116.19542911 -107.30467024]

I know though that the first results is correct. I assume that I am passing incorrectly the translation and rotation matrixes and start to think that maybe I need to actually pass the inverse rotation and translation. Could you maybe check what relationship exists between the parameters of opencv triangulate points and your input? Thanks a lot!

No worries. Still I was not able to pass a vector as column vector. Even if I do, it becomes a row vector, so you should check in the beginning and transpose it if necessary. Now I am checking the result of the function to see if I get the expected output :)

I finally made it almost work. I noticed that you are doing this: CV_Assert( ( tvecs.size() == pts.checkVector( 2, CV_32F, true ) ) ); assuming that the vector contains float32, but then almost all computation is done using double, so the function should accept also float64

Ok thanks, that helped :) Now I got here : cameraTranslation = (-cameraRotation * tvec); but the sizes don't match. cameraRotation is 3x3 but tvec is 1x3. I could transpose tvec but its not correct. Probably the tvecs should have another shape. What shape should they have in the input function calcPosition() ?

I am almost there but I get "calibration.cpp:292: error: (-201) Input matrix must be 1x3, 3x1 or 3x3 in function cvRodrigues2" for the rotation matrix vector. I am passing a numpy array of rotation matrixes (numpy matrixes of size 3x3), but it seems like it is expecting something different. The problem seems to be that I am passing an array of 33 elements, each on of size 3x3, but once passed to calcPosition, if I print the size of _rvecs it is [3 x 33]

I actually compiled your fork and would like to know how to call it from python. Could you help me with that?

Hi! Since I would really need this implemented algorithm, do you have a stand alone version which is not integrated in opencv which I could use without waiting for the merging?

Thanks a lot :)

Great! also looking forward to having this. Are you going to create also a Python interface?

Hi, I have a 512x424 CV16U image which is the depth data of the kinect2. I want to resize it to 640x480 interpolating it. I tried with cv::resize(depth, depth_resized, cv::Size(640,480)) but the image becomes black :( Just to check if everything is correct I copied the original image in a bigger mat (640x480) and it works depth.copyTo(depth_resized(cv::Rect(64, 28, 512, 424)));

I think that the interpolation process is interpolating the data as if it was 8bit instead of 16. Thanks