Here's how to calibrate your camera!
Note: In the process Point Cloud menu, you'll see the option to specify the output, including naming your project and the Laz file. You can also select the optimization level. Once done, locate the start processing button. To its right is a downward arrow which leads you to the camera calibration process.Understanding the camera calibration process is crucial for users of the Rock R3 Pro. This LiDAR system provides the flexibility of removing the camera for handheld SLAM dock use, before reattaching it for drone-based data collection. However, this necessitates the need for precise camera calibration to achieve optimal alignment between the RGB and the LiDAR data in intensity view. This article will provide an in-depth, step-by-step guide to camera calibration using Rock Desktop.
1. Preparing for Calibration
Start your calibration process by opening the Rock Desktop and follow the steps to produce a Point Cloud. These steps include processing the trajectory, selecting the trajectory, manipulating the angle gate and the range gate. Once completed, navigate to the process Point Cloud menu.
Note: In the process Point Cloud menu, you'll see the option to specify the output, including naming your project and the Laz file. You can also select the optimization level. Once done, locate the start processing button. To its right is a downward arrow which leads you to the camera calibration process.
2. Choosing the Ideal Photos for Calibration
A key element to a successful camera calibration is choosing good photos. These photos should feature distinct characteristics in all corners and the center. Ideal images should contain varied features across their entire surface area. Photos of parking lots with clear lane markings are great examples.
Tip: To obtain a quality dataset for calibration, fly the drone at approximately 40m, moving slowly at about 2 m/s, with a 50% overlap. This ensures a sufficient data density in the Point Cloud to find good edges and corners.
Once you've selected your photos, click Next.
3. Adding Key Pair Point Matches
Rock Desktop will now display a single photo of the Point Cloud data. Your task is to add key pair point matches in both the intensity view and the RGB view.
- Click on the plus button to add an RGB point.
- Add an intensity viewpoint corresponding to the RGB point's location.
You should add multiple points on the ground in a single region to offset any potential error in point selection.
Important: Ideally, you should select about five points each in the top left, top right, bottom left, bottom right corners, and across the center. You don't have to do this all in one image. You can cycle between the selected images by clicking 'Next' at the bottom and continue to add points across images.
To aid in identifying matches between the intensity view and the RGB view, adjust the brightness of the intensity view using the slider. This will help reveal the features and edges you need to see.
4. Calibrating the Camera
Once you've selected all the key pair point matches, click 'Next'. You'll be presented with three options:
- Calibrate the Intrinsics
- Calibrate the Extrinsics
- Calibrate both the Intrinsics and Extrinsics
Important: You should only calibrate the extrinsics on the R3 pro. Intrinsics are fixed to the camera and change minimally over time. The camera's orientation with respect to the LiDAR body, however, changes whenever you attach or detach the camera. This could result in minute movements even if the camera is tightly fastened. Therefore, to get a precise calibration for the R3 Pro, just select 'Extrinsics'. For R2A and R360 you'll need to do Extrinsic and Intrinsic.
Following this comprehensive guide will ensure you obtain the best results from your Rock R3 Pro system. Precise camera calibration is essential for achieving optimal alignment of RGB and LiDAR data, which ultimately leads to higher quality, more accurate outputs.
Remember: Regular calibration is necessary whenever the camera is detached and reattached to the LiDAR system. Always check calibration before embarking on a data collection project to ensure the most accurate results possible.