KITTI Coordinate Transformations

Original Source Here

LiDAR-Camera 3D bounding-box Transformation

The steps mentioned below defines the procedure of converting a 3D bounding box from LiDAR coordinate frame to camera coordinate frame.

  1. 3D BB centroid (x, y, z) projection from LiDAR coordinate frame to camera coordinate frame

a. Rigid body transformation

b. Rectification handling

2. Heading/yaw angle transformation

Heading/yaw angle is w.r.t z-axis in the LiDAR coordinate frame, which corresponds to -(y-axis) in camera coordinate frame

3. 3D bounding box corner coordinates calculation (camera coordinate frame)

Figure 3: 3D Bounding Box: LiDAR coordinate system

x-axis -> right (length), y-axis -> bottom (height), z-axis -> forward (width)
3B bb corner coordinates in camera coordinate frame, coordinate system is at the bottom center of the box.

4. Rotation of the 3D BB corner coordinated by rotation_y angle (camera coordinate frame)

5. Box translation by centroid coordinates (camera coordinate frame)


Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot

%d bloggers like this: