I have got the following camera parameters that I am trying to integrate into the code I have regarding the task shown in the image; Camera $1($Frame$1)$:

Intrinsic matrix:

$[[1\mathrm{.}02408777$e$+03,0.,6\mathrm{.}01806702$e$+02],$

$[0.,1\mathrm{.}02389478$e$+03,5\mathrm{.}08131683$e$+02],$

$[0.,0.,1.]]$

Distortion: $[-2\mathrm{.}52257544$e$-03,4\mathrm{.}38565714$e$-03,0.,0.,9\mathrm{.}21500759$e$-05]$

Camera $2($Frame$2)$:

Intrinsic matrix:

$[[1\mathrm{.}02419836$e$+03,0.,6\mathrm{.}96750427$e$+02],$

$[0.,1\mathrm{.}02398749$e$+03,5\mathrm{.}07494263$e$+02],$

$[0.,0.,1.]]$

Distortion: $[-3\mathrm{.}26306466$e$-03,5\mathrm{.}70008671$e$-03,0.,0.,7\mathrm{.}57322850$e$-05]$

Transformation between cameras $($from Camera $1$ to $2)$:

R$\_$vec:

$[[9\mathrm{.}99999404$e$-01,-1\mathrm{.}24090354$e$-06,-1\mathrm{.}11142127$e$-03],$

$[1\mathrm{.}29263049$e$-06,1.,4\mathrm{.}65405355$e$-05],$

$[1\mathrm{.}11142127$e$-03,-4\mathrm{.}65419434$e$-05,9\mathrm{.}99999404$e$-01]]$

T$\_$vec $=[[-4\mathrm{.}34818459$e$+00],$

$[2\mathrm{.}83603016$e$-02],$

$[-9\mathrm{.}00963729$e$-04]][$mm$]$

The distortion coefficient vectors are defined as $[$k$1,$ k$2,$ p$1,$ p$2,$ k$3]$ where kn are the radial

distortion coefficients and pn are the tangential distortion coefficients. and the code is as below: import cv$2$

import numpy as np

import matplotlib.pyplot as plt

# Paths to the images

left$\_$image$\_$path $=$ 'left$\_$frame.png$\text{'}$

right$\_$image$\_$path $=$ 'right$\_$frame.png$\text{'}$

# Function to safely load images and resize if they don't match

def load$\_$and$\_$resize$\_$images$($left$\_$image$\_$path, right$\_$image$\_$path$)$:

left$\_$image $=$ cv$2.$imread$($left$\_$image$\_$path, cv$2.$IMREAD$\_$GRAYSCALE$)$

right$\_$image $=$ cv$2.$imread$($right$\_$image$\_$path, cv$2.$IMREAD$\_$GRAYSCALE$)$

if left$\_$image is None or right$\_$image is None:

print$("$Error loading images."$)$

return None, None

try:

# Resize the images to a common size if they don't match

if left$\_$image.shape $!=$ right$\_$image.shape:

common$\_$size $=($min$($left$\_$image.shape$[1],$ right$\_$image.shape$[1]),$

min$($left$\_$image.shape$[0],$ right$\_$image.shape$[0]))$

left$\_$image $=$ cv$2.$resize$($left$\_$image, common$\_$size$)$

right$\_$image $=$ cv$2.$resize$($right$\_$image, common$\_$size$)$

except Exception as e:

print$($f$"$Error resizing images: $\{$e$\}")$

# Return None to indicate an issue

return None, None

return left$\_$image, right$\_$image

# Paths to the images

left$\_$image$\_$path $=$ 'left$\_$frame.png$\text{'}$

right$\_$image$\_$path $=$ 'right$\_$frame.png$\text{'}$

# Load and resize images

left$\_$image, right$\_$image $=$ load$\_$and$\_$resize$\_$images$($left$\_$image$\_$path, right$\_$image$\_$path$)$

# Proceed only if both images are successfully loaded and resized

if left$\_$image is not None and right$\_$image is not None:

# Continue with your processing here...

# b$.$ Find Corresponding Features and Create Composite Image

orb $=$ cv$2.$ORB$\_$create$()$

keypoints$1,$ descriptors$1=$ orb.detectAndCompute$($left$\_$image, None$)$

keypoints$2,$ descriptors$2=$ orb.detectAndCompute$($right$\_$image, None$)$

bf $=$ cv$2.$BFMatcher$($cv$2.$NORM$\_$HAMMING, crossCheck$=$True$)$

matches $=$ bf$.$match$($descriptors$1,$ descriptors$2)$

matches $=$ sorted$($matches$,$ key$=$lambda x: x$.$distance$)$

composite$\_$image $=$ cv$2.$drawMatches$($left$\_$image, keypoints$1,$ right$\_$image, keypoints$2,$ matches$[$:$10],$ None$)$

plt$.$imshow$($composite$\_$image$),$ plt$.$show$()$

# c$.$ Estimate Fundamental Matrix

points$1=$ np$.$float$32([$keypoints$1[$m$.$queryIdx$].$pt for m in matches$]).$reshape$(-1,1,2)$

points$2=$ np$.$float$32([$keypoints$2[$m$.$trainIdx$].$pt for m in matches$]).$reshape$(-1,1,2)$

fundamental$\_$matrix, mask $=$ cv$2.$findFundamentalMat$($points$1,$ points$2,$ cv$2.$FM$\_$RANSAC$)$

print$($fundamental$\_$matrix$)$

# d$.$ Find Correctly Matched Points with Epipolar Constraint

correct$\_$matches $=[$m for m$,$ msk in zip$($matches$,$ mask$)$ if msk$[0]==1]$

# e$.$ Estimate Shaft Diameter using Disparity Map or $3$D Reconstruction

# Implement stereo vision algorithms and $3$D reconstruction techniques here

stereo $=$ cv$2.$StereoSGBM$\_$create$($minDisparity$=0,$ numDisparities$=16,$ blockSize$=5)$

disparity$\_$map $=$ stereo.compute$($left$\_$image, right$\_$image$)$

# Extract relevant region and calculate diameter based on disparity

relevant$\_$region $=$ disparity$\_$map$[150$:$300,200$:$400]$ # Adjust region based on your dataset

shaft$\_$diameter $=$ np$.$max$($relevant$\_$region$)$ # Assuming the shaft is the highest disparity value

print$("$Estimated Shaft Diameter:", shaft$\_$diameter$)$

# Display disparity map

plt$.$imshow$($disparity$\_$map, cmap$=$'plasma'$),$ plt$.$show$()$

else:

print$("$Skipping image pair due to an error."$)$