How i can retrieve 3D position of points from 2D images of an object?

I have two images from a car which are taken from different pov. the objective is to retrieve the 3D position of the symmetric features with respect to the car symmetry plane and reconstruct the car symmetry plane (which is the distance between axes and the tiers). I have two vanishing points from the conic plane of the wheels and another vanishing point from the symmetric feature (license plate) for both images. I just attached one image to give more clues.

i retrieved the camera calibration (K) matrix for both images. my question is since i don't have any information about the internal camera calibration, can i assume that the reconstruction is an affine reconstruction? and another question can i use the vanishing points from the second image to build the 3D position of the point? if yes how? because i know i need the F matrice and H matrices, to be able to retrieve the 3D position of symmetric features, but im not sure which method is better to use for estimating the F and H matrices.

Answer 1

[Pratyush Swain]

Hi Homeyra,

I understand you want to retrieve the 3D position of points from 2D images of an object.Please refer to this example : https://www.mathworks.com/help/vision/ug/structure-from-motion-from-multiple-views.html which illustrates the process of estimating the 3-D structure of a scene from a set of 2-D images.

You can also refer to https://www.mathworks.com/help/vision/ref/cameracalibrator-app.html app which helps to estimate camera intrinsic and extrinsic parameters. With regard to fundamental(F) matrix , please refer to https://www.mathworks.com/help/vision/ref/estimatefundamentalmatrix.html to estimate fundamental matrix from corresponding points in stereo images.

Hope this helps.