I think the fused point cloud is very nice, but resolution is not high.
I thought following method to get image corresponding to points in the fused point cloud.
Handle pre-recorded svofiles.
- Obtain images corresponding to chunk from chunk.timestamp and timestamp when request_spatial_map_async is executed
- Use OpenCV projectPoints to get the u,v of the image from the chunk vertices
However, u and v obtained by this method are far from ideal values.
The parameters used for projectPoints are required as follows.
py_rotation=sl.Rotation() zed_pose.get_rotation_matrix(py_rotation) rvec=py_rotation.get_rotation_vector() py_translation = sl.Translation() tvec=zed_pose.get_translation(py_translation).get() cameraMatrix = np.array([[camera_parameters.fx,0,camera_parameters.cx], [0,camera_parameters.fy,camera_parameters.cy],[0,0,1]])
Please give me advice