# Revision history [back]

Given you're written the majority of the code for this it's only a tiny bit of work to get this working. Just to give a bit of background, there are two different types of point cloud that can be represented by PCL: Dense structured point clouds and sparse point clouds.

In dense point clouds, the points are stored in a grid structure a NxM matrix of 3D points with invalid points being represented as vectors of NANs. These types of point clouds are generally created by depth cameras and stereo vision algorithms.

In sparse point clouds, the points are represented as a flat list of points in no specific order these are more commonly generated by LIDAR sensors.

Getting back to your question, the kinect produces dense structured point clouds so you will be able to 'remove' a point by simply setting its x, y and z values to NAN. This means that the grid structure of the points is preserved.

Hope this helps.

Given you're written the majority of the code for this it's only a tiny bit of work to get this working. Just to give a bit of background, there are two different types of point cloud that can be represented by PCL: Dense structured point clouds and sparse point clouds.

In dense point clouds, the points are stored in a grid structure a NxM matrix of 3D points with invalid points being represented as vectors of NANs. These types of point clouds are generally created by depth cameras and stereo vision algorithms.

In sparse point clouds, the points are represented as a flat list of points in no specific order these are more commonly generated by LIDAR sensors.

Getting back to your question, the kinect produces dense structured point clouds so you will be able to 'remove' a point by simply setting its x, y and z values to NAN. This means that the grid structure of the points is preserved.

Hope this helps.

Update

You can set the values of a point using the iterator you are using in your loop as shown below:

it[0] = std::numeric_limits<float>::quiet_NaN();
it[1] = std::numeric_limits<float>::quiet_NaN();
it[2] = std::numeric_limits<float>::quiet_NaN();


The iterator functions like a pointer to the pcl::Point allowing you to get and set its values.