Comparing two point clouds for differences

I am attempting to use two point clouds from the kinect and compare for differences between the two scenes.

The logic is as follows: For each point in point cloud 1, the distance to the nearest neighbor in point cloud 2 is calculated. If the distance is greater then the threshold distance, the indices of the point in point cloud 1 is stored. After completing for loop, the indices are used to select the points in point cloud 1 that were greater than threshold difference.

function [PCDifference] = ComparePC(pc1,pc2)
%Compares two aligned point clouds based on threshold difference
[pc1r,~]=removeInvalidPoints(pc1);
[pc2r,~]=removeInvalidPoints(pc2);
minDist = 0.5;

x=0;

%for each point in pc1 find nearest neighbor distance - if greater then
%threshold distance store point cloud in PCDifference
for i =1:pc1r.Count
    point=pc1r.Location(i,:);
    [~,dist]=findNearestNeighbors(pc2r,point,1);
    if dist > minDist

    if true
      % code
    end
          %Store point cloud in indices
          x=x+1;
          Indices(x,1)=i
      end
  end
  %Plot difference in point clouds
  PCDifference=select(pc1r,Indices);
  figure
  pcshow(PCDifference)

I am receiving errors in PCDifference=select(pc1r,Indices) line. Indices should be a column vector with the indices of points without close neighbor. Is there a better way to approach this? There is no easy way to create point cloud object that changes size every loop. It is best to store the indices and then use the select function afterwards.

Please let me know if you have any ideas.

Thank you, Michael

Answer 1

[KSSV]

Doc _ knnsearch_ ...with this you can get the indices of nearest neighbors with ease..

[Prashik Shende]

can use specify how to use knnsearch in code