Calculating D1(point to point) and D2(point to plane) Geometry PSNR for point clouds in MATLAB

I am working on the 3D point cloud. could you please guide me on how I can calculate D1(point to point) and D2(point to plane) Geometry PSNR in MATLAB

Answer 1

[Maneet Kaur Bagga]

Hi Faranak,

• As per my understanding to calculate D1 and D2 Geometry PSNR in MATLAB, you need to calculate the Euclidean distance between corresponding points in the two point clouds for D1.
• For D2, fit planes to the point clouds and measure the perpendicular distance from each point to its corresponding plane. Then, calculate the mean squared error (MSE) between the original and reconstructed distances and use it to calculate the PSNR.

Please refer to the code below to calculate D1:

% Number of points in clouds A and B
nA = 10;
nB = 15;

% Generating random coordinates for clouds
ptsA = randn(nA, 3) * 10;
ptsB = randn(nB, 3) * 5;

% Point Clouds A and B
ptCldA = pointCloud(ptsA);
ptCldB = pointCloud(ptsB);

% pcshow(ptCld1)
% Implementing D1 PSNR for point cloud A

D1_AB = 0;

for i = 1:nA
    ptA = ptCldA.Location(i, :);
    % Nearest point to the ith point A in point cloud B
    [nearestPtB, ~] = findNearestNeighbors(ptCldB, ptA, 1);
    % Square of l2 norm
    d = norm(ptA - ptCldB.Location(nearestPtB, :)) ^ 2;
    D1_AB = D1_AB + d;
end

D1_AB = D1_AB / nA;
disp(D1_AB);

   61.0346

Please refer to the following documentation for more information:

Point Cloud Processing

https://www.mathworks.com/help/vision/point-cloud-processing.html

pointCloud function

https://www.mathworks.com/help/vision/ref/pointcloud.html

findNearestNeighbors

https://www.mathworks.com/help/vision/ref/pointcloud.findnearestneighbors.html

Hope this helps!

Regards,

Maneet Bagga