SVIn–Perdix Matcher

Synchronizes an SVIn trajectory with a Perdix dive-computer export and calibrates the SVIn depth measurements against the Perdix depth reference.

The script finds the time offset between the recordings, interpolates the SVIn trajectories at the Perdix sample times, fits a linear depth correction, and writes the corrected SVIn data and diagnostic plots to an output directory.

Requirements

• Python 3.9 or newer
• NumPy
• pandas
• Matplotlib
• SciPy
• scikit-learn
• Open3D

Installation

To install Python dependencies and set up the workspace, run:

mkdir ~/depth_matcher_ws
cd ~/depth_matcher_ws
git clone https://github.com/AutonomousFieldRoboticsLab/Svin-Perdix-Matcher .
sudo apt install python3-venv
python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install -r requirements.txt

Input files

Perdix CSV

Use the CSV exported by the Perdix dive computer. The matcher expects the export's metadata rows and these columns:

• Dive Number for elapsed time and its unit
• GF Minimum for depth samples
• Start Date and End Date for recording bounds

The depth unit must be supplied separately as either m or ft.

SVIn text file

The SVIn file must be whitespace-delimited and contain:

• #timestamp with Unix timestamps in seconds
• tz with depth measurements

Optional point cloud

If you have a .ply point cloud that should be shifted by the same calibrated depth offset, pass it with --pointcloud.

Usage

python svin_perdix_matcher.py PERDIX_CSV SVIN_TXT OUTPUT_DIRECTORY {m,ft} [--pointcloud POINT_CLOUD.ply]

Example using the bundled data:

python svin_perdix_matcher.py \
  data/CatacombsPerdix.csv \
  data/svin_example.txt \
  output \
  ft \
  --pointcloud data/pointcloud_example.ply

This creates:

• output/svin_example_matched.txt
• output/depth_offset.txt
• output/pointcloud_example_matched.ply (only when --pointcloud is used)
• several diagnostic PNG files in output/

Example outputs

Original recordings	Time-aligned and calibrated result

Output details

The main output is the matched SVIn TXT file:

• *_matched.txt
• same original whitespace-delimited SVIn layout
• tz values are replaced with the calibrated depth values

Additional diagram outputs in output/ include:

• original_data.png
• shifted.png
• interpolate.png
• regression.png
• final.png
• depth_offset.txt (saved calibration offset)
• *_matched.ply when a point cloud is passed

Notes

• Perdix timestamps are interpreted as UTC.
• The Perdix recording must be at least as long as the SVIn recording.
• The time offset is searched at 0.1-second resolution.
• Existing output files are overwritten.
• The depth offset is stored in depth_offset.txt and can be reused for aligned point-cloud processing.