For those of you that process mobile LIDAR data, it can be necessary to filter out the points that represent moving cars. This can be accomplished using the Classify Moving Objects routine. This tool classifies groups of points that are captured from moving objects. It can be used for mobile laser data sets. The routine requires that data from multiple lines or scanners is available. The routine classifies groups of points that are captured at one moment of time and at another moment of time, when the area was captured by another scanner or in another line, there are no points at the same location, but points from behind the moving object.
Classifying points that represent cars requires that points have been assigned to groups. This can be done using the Assign Groups command which assigns a group number to points of one or more classes. It is best to use only “Group by density” option and to apply it to points high enough to avoid using noisy ground data.
In order for group numbers to be assigned to points, Compute normal vectors command should be used. This command can be used to compute and store two additional attributes for laser points, a dimension and a normal vector. It is strongly recommended to have trajectory information available for computing normal vectors for mobile ground-based laser data.
Below are some general steps to this workflow:
- Have compatible LIDAR Data and Trajectory information
- Execute Compute normal vectors command
- Execute Compute Distance command using the “from ground” option.
- Execute Assign Groups command
- Execute Classify Moving Objects routine