Noise detection

The noise detection classifies points which are isolated in the point cloud as points of the class 7, which is the class for low points (noise) in the ASPRS standards.

It is important to note that this detection will only find isolated points: points which are distant from any other points in the project. The algorithm will not be able to detect noise created by moving objects, such as moving cars, because of the higher point count. Surface noise will not be detected either because of the proximity of other structures.

Different settings are available for this function:

• Min. points per cell: Number of points in a cell for the cell to be considered as noise by the algorithm.

• Minimum cell size (in distance units): Minimal size of tiles used in this algorithm. This value determines the precision of the detection. The smaller the value, the more precise the ground will be, but the longer the process will take. The minimal size of the cell should be smaller than the size of a project cell, or else the size of the project cell will be used. The information about the project’s cell size can be found in the menu “File / Project info”.

• Starting point: Point where the algorithm will begin its process. It can be manually set or picked in the point cloud with the “Pick” button. Ideally a distant point is selected in order to minimise processing time.  If “Limit by radius” is used, the detection will be centered around this point.

• Limit by radius (in distance units): Limits the zone to be processed around the starting point. This option is useful when trying the best combination of parameters in a project or for small areas which are distinct from the rest of the project.

• Use statistical removal: This option will consider the position of nearby points in neighboring cells. The advantage of using this option is that cells with a few points at their extremity but which are part of a larger group of points will not be marked as noise. However, this option adds processing time to the detection.