Plane-measument

This repo offers a way to position and measure distances and areas from a RGB photo from a 2d camera.

USAGE

1. Calibrate a camera with opencv-python. You can use: cv2.cameraCalibration() to get the camera matrix an the distortion coeficients.
2. Use the script: get_position_maps.py as follows:

python3 get_position_maps.py 
    --cfg_path '/path/to/camera_calibration_pickles/' 
    --height 12 
    --angle 45.0 
    --roll 5.3 
    --imgshape 3000 4000

Now you have:

• Matrix with the same shape as the original images that stands for the $m^2/px^2$ convertion in every pixel
• Two other matrix with the same shape as the original images that stand for the $x$ and $y$ position of every image's pixel.

Results

• Position:

• Area

(A4 sheet area is $0.06237\ \mathrm{m^2}$)

Some plot from calculus

- Intrinsic camera values:

• Pixel FoV (FIeld of View)

- Extrinsic camera values for a given configuration (dependent on intrinsic parameters):

• Position maps

• Area map:

• Visual solution:

References:

A_Method_for_Measuring_3D_Distances_from_a_2D_Image_Without_Knowing_Depth_or_Having_a_Known_Pattern. DOI: 10.13140/RG.2.2.34832.38401