High-Precision Motion Compensation for LiDAR Based on LiDAR Odometry

Figure 5

Filtering effects of key points in scenarios with different thresholds. (a–d) The key point filtering results when is 5, 10, 15, and 20. It can be seen from the figure that when is 5, the distribution of key points in the scene is still redundant; when , the value of key points is too sparse, which is not conducive to subsequent feature point matching.

(a)

(b)

(c)

(d)