… In this work, we propose to align semantic features instead of edge features to improve
online calibration robustness, especially for low-resolution LiDAR and noisy inputs. …

… To evaluate the importance of semantic edge features 7 in our work, we replace semantic
edge features on point clouds 8 with geometric edge features detected by a lightweight …

… Due to the wide availability of edge features in natural indoor and outdoor scenes, our
method aligns these edge features observed by both LiDAR and camera sensors. Fig. 2 further …

… features consist of the 3D depth edges on both sides of the target and the 3D corner in the
center. Image features are intensity edges extracted from both edges … 3D features projected on …

… a new method for automatic extrinsic calibration between camera and LiDAR sensors. We
based our proposed approach on edge feature camera-LiDAR co-registration in arbitrary …

… LiDAR and camera edge features, without distinguishing whether the LiDAR and camera
features … However, it should be noted that when only a few LiDAR and image frames are used …

… We track the extracted edge and surf features in the current LiDAR scan back to the … find
two edge features in the old scan, Ll pfj, Ll pfk, corresponding to the projected edge feature Ll pfi. …

… complementary measurements of a camera and LIDAR, we need to … camera-LIDAR
sensor system is to extract the geometric feature from both measurements and register the “feature…

… observe the same features, even if their perspectives are nearly identical. In this section,
we focus on promoting edge features from camera images. Edges are located where pixels …

… operator is used to extract the edge from the image. From the assumption that the depth of
the lidar point cloud corresponds to the edge feature, the lidar edge feature map E L can be …