In phyton Implement the photometric stereo technique

Implement the photometric stereo technique described in the lecture. Your program must have two

parts:

Given the images and corresponding light source directions, estimate the surface normals and

albedo map.

Reconstruct the depth map from the surface with the implementation of the Horn integration

technique given above in horn$\_$integrate function. Note that you will typically want to run

the horn$\_$integrate function with $10000-100000$ iterations, meaning it will take a while.

$[22]$: def photometric$\_$stereo$($images$,$ lights, mask, horn$\_$niter$=25000)$:

$"$n$"$

inputs :

images : $(\{$:n$\_(-)$ims$,$h$,$w$)$

lights : $($n$\_$ims, $3)$

mask: $($h$,$w$)$ an optional parameter which you are encouraged to use.

It can be used e$.$g$.$ to ignore the background when integrating the

$-$normals.

It must be created by converting the images to grayscale and,

~thresholding

$($only using locations for which the pixel value is above some $\}$

$-$threshold$).$

The choice of threshold is something you can experiment with,

but in practice something like $0\mathrm{.}05$ tends to work well.

$"""$

Write your code here.

albedo $=$ np$.$ones$($images$[0].$shape$)$

normals $=$ np$.$dstack$(($np$.$zeros$($images$[0].$shape$),$

np$.$zeros$($images $[0].$shape$),$

np$.$ones$($images$[0].$shape$)))$

H$\_$horn $=[]$

return albedo, normals, H$\_$horn