- First you create an ‘interpolant’ using the [x, y] coordinates where valid distance (z) value is available.
- Create query point/s ‘[xq, yq]’ and use the interpolant object from step 1 to get ‘zq’ (the expected distance) corresponding to the query points.
Fill the occlusion in 3D point cloud data
5 次查看（过去 30 天）
Currently am working in calcualting volume for the point cloud data. Am using depth data 'z' as an important parameter for volume calculation. Due to occlusion some point doesnt have z value and the corresponding x and y values also will not be there. I will attach the ss for this case. My need is to fill the occluded region with its neighbour (can be mean, min or max).
I assume that there are invalid points in the point cloud you generated. Invalid points will be of form ‘[Inf, Inf, Inf]’ or ‘[NaN, NaN, NaN]’. I suggest you use ‘2D interpolation’ from scatteredInterpolant.
It works in two stages:
Here '[xq, yq]' points are to be decided by the user.
You may also want to refer to following resource on 3D point cloud interpolation -