[pcl/registration/ia_ransac.h]In the function findSimilarFeatures(), the statement feature_tree_->nearestKSearch(input_features, sample_indices[i], k_correspondences_, nn_indices, nn_distances) performs nearest neighbor feature search. I cannot locate the specific implementation code for this. I would like to view this code and would appreciate any assistance.

If i have a point cloud, for example of a metal object that has a few extrusions/unwanted parts like in the image, are there any clustering methods to remove such extrusions? I’ve tried using Euclidean Cluster Extraction from PCL, but as expected it cant isolate them very well. So i was wondering if there are other methods that i could use. If clustering is not the right method at all, what else can i look into?

eimerI have a curved object as a point cloud and using pcl, i first downsampled it using voxelgrid and then cluster extracted my object of interest and then used normal estimation to find normal vectors.
If my task is to find the normals equidistant from each other of a distance of 10mm, how would i go about it? Basically i need to reduce the number of normal vectors (Downsampled data size is 6000+ and so is the normal vector data size the same). i tried changing the leaf size of voxelgrid but then it doesnt look right.