With the help of mobile laser scanners it is possible to create three dimensional city models quickly and precisely. These city models can be used to create city development plans, to animate movies or to develop computer games. However during the scanning process, cars moving and pedestrians walking the streets can cause noise which is decreasing the quality of the output point cloud. To create as precise city models as possible, the removal of these noises is mandatory.
In my BSc theses I provide a method to filter these noises. The method is based on the segmentation of the input point cloud multiple times with different parameters. In each iteration it is classifying the objects to static objects or possible noises by different properties of the segments. The output of the algorithm is a point cloud containing the points that are static objects for sure. Furthermore I demonstrate the necessary tools and algorithms to manipulate three dimensional point clouds and those that are my method based on. And I also demonstrate the program and the usage of the program that implements this method.
The provided solution with the right parameters is able to remove all the noise generated by moving objects from any point clouds acquired by mobile laser scanners. The drawback of the algorithm is that it will classify the poor quality parts of the input cloud as noise as well.