Segmentation Fault while doing initial_guess_manual on Ouster sample dataset

[devvaibhav455]

Before opening an issue
If the issue is about build errors:

• I have read installation instructions
• I have considered using docker It gave me build errors while doing on arm64

If the issue is about calibration procedure and results:

• [NA] I have read data collection instructions
• [NA] I have confirmed the program works on example data

If the issue is about the algorithm:

• [NA] I have read the paper

Describe the bug
Segmentation fault while doing the initial guess

To Reproduce
Steps to reproduce the behavior:

1. export LD_PRELOAD=/usr/lib/aarch64-linux-gnu/libopencv_core.so.4.5.4d as there was a warning during the workspace build process

--- stderr: direct_visual_lidar_calibration                                          
** WARNING ** io features related to pcap will be disabled
/usr/bin/ld: warning: libopencv_core.so.4.5d, needed by /opt/ros/humble/lib/libcv_bridge.so, may conflict with libopencv_core.so.408
---

and below while preprocessing if I don't export the LD_PRELOAD variable

terminate called after throwing an instance of 'cv::Exception'
  what():  OpenCV(4.8.0) /home/ubuntu/opencv_build/opencv/modules/core/src/matrix.cpp:246: error: (-215:Assertion failed) s >= 0 in function 'setSize'

[ros2run]: Aborted

3. Preprocess the data (successful)
4. Ran find_matches_superglue.py and initial_guess_auto successfully
5. Run initial_guess_manual

ros2 run direct_visual_lidar_calibration initial_guess_manual ~/Downloads/temp/ouster_preprocessed/
camera_model:plumb_bob
intrinsics  :4
distortion  :5
loading /home/nvidia/Downloads/temp/ouster_preprocessed//rosbag2_2023_03_28-16_25_54.(png|ply)
loading /home/nvidia/Downloads/temp/ouster_preprocessed//rosbag2_2023_03_28-16_26_51.(png|ply)
image_size:2448x2048
camera_fov:57.1854[deg]
Gtk-Message: 23:13:30.941: Failed to load module "canberra-gtk-module"
[ros2run]: Segmentation fault

So, I installed canberra-gtk-module using sudo apt install libcanberra-gtk-module libcanberra-gtk3-module (successful)

6. Re-run

ros2 run direct_visual_lidar_calibration initial_guess_manual ~/Downloads/temp/ouster_preprocessed/
camera_model:plumb_bob
intrinsics  :4
distortion  :5
loading /home/nvidia/Downloads/temp/ouster_preprocessed//rosbag2_2023_03_28-16_25_54.(png|ply)
loading /home/nvidia/Downloads/temp/ouster_preprocessed//rosbag2_2023_03_28-16_26_51.(png|ply)
image_size:2448x2048
camera_fov:57.1854[deg]
[ros2run]: Segmentation fault

Expected behavior
Initial guess manual should run without errors on the sample dataset

Environment:
Ubuntu 22.04 on Jetson Jetpack 6.x (arm64)
ROS version: ROS2 Humble

Additional context
I am using the e47a42c2957951c9fafcca9995d9927e15557069 ceres_solver version as mentioned in #114
Using VNC to access jetson