Figure 1.
Figure 2.
Figure 3.
Figure 4.
COMPARISON OF VARIOUS POINT CLOUD SEGMENTATION METHODS
| segmentation methods | Advantage | Disadvantage |
|---|---|---|
| edge-based methods | Can detect the edges of different areas very intuitively for point cloud. | sensitive to noise and not suitable for objects with smooth surface changes. |
| region-based methods | More accurate than edge-based methods. | The segmentation result depends on the quality of the seeds and the merging rules. There will be over-segmentation or under-segmentation. |
| model-based methods | Fast segmentation speed, and heterogeneous,suitable for simple geometric models. | Difficult to use in complex scenarios. |
| graph-based methods | Suitable for complex scenes. | Lack of real-time. |
| machine learning-based methods. | Point cloud segmentation has high accuracy, good recognition effect. | lack of real-time. |