/MosaicSeadron

Primary LanguagePython

SeaDroneLib

MosaicSeadron is a repository that includes all the necessary scripts to georeference, mosaic, and fully map consecutive overlapped captures taken by Unmanned Aerial Vehicles (UAVs, or drones) equipped with RGB, thermal, and multispectral sensors. More information about the methods employed in this repository can be found in "Enhancing Georeferencing and Mosaicking Techniques over Water Surfaces with High-Resolution Unmanned Aerial Vehicle (UAV) Imagery", a research article authored by Alejandro Román, Sergio Heredia, Anna E. Windle, Antonio Tovar Sánchez, and Gabriel Navarro. The research article is published in: https://www.mdpi.com/2072-4292/16/2/290

The code inside gathers features like:

  • Extract camera metadata of Micasense sensors.
  • Extract camera metadata of DJI sensors.
  • Georeference individual UAV captures.
  • Merge (mosaic) consecutive UAV captures.
  • Filter captures and/or merges.

Folder Structure

Folder Structure

In general there will always be the following folders:

  • dependencies:
    • contains extra files that are necessary to run the code.
  • micasense:
    • code to process the screenshots taken with the micasense camera
  • scripts:
    • files that allow to execute configuration functionalities such as installing or uninstalling the micasense code
  • seadrone:
    • code for geo-referencing and mosaicking
  • seadrone_usage:
    • codes that collect typical use cases such as rrs extraction, georeferencing, ...

Set Up

Here we explain how to download and install the code to be able to work with it.

Clone the project

  git clone https://github.com/Aouei/seadronelib

Go to the project directory

  cd seadronelib

Important: The environment must have python3.10 to work.

Create a virtual environment with venv on Windows

  python3.10 -m venv seadronelib-venv --prompt="seadronelib"

Activate the environment with venv

  seadronelib-venv\Scripts\activate

Create a virtual environment with anaconda on Windows

  conda create -n seadronelib python=3.10 -y

Activate the environment with anaconda

  conda activate seadronelib

Install the dependencies

  python -m pip install -r requirements.txt
  python pip install dependencies\GDAL-3.4.3-cp310-cp310-win_amd64.whl

Install micasense and seadrone modules

Important: The virtual environment must be activated

Generation of install.sh file and install micasense and seadrone

  python -B .\scripts\package_manager_generator.py -e seadronelib-venv -p micasense,seadrone -i 1 -ri 1

Generation of uninstall.sh file once micasense and seadrone were installed

  python -B .\scripts\package_manager_generator.py -e seadronelib-venv -p micasense,seadrone -u 1

If you want to install micasense and seadrone

  \scripts\install.sh

If you want to uninstall micasense and seadrone

  \scripts\uninstall.sh

Technical Description

The code is mainly composed of 4 files:

  • raster.py:
    • contains the code related to georeference, mosaicking, etc...
  • data_structures.py:
    • contains classes for gathering useful data needed in the processing.
  • enums.py:
    • contains enums to summarize some useful data like sensors available, etc...
  • processing.py:
    • contains the code to process a single flight in a simpler and user friendly manner.

Class Diagram

Each file (folder symbol) contains a set of classes (rectangles) with methods to solve the needs we have such as extract rrs, georeference, etc...

Each method has its own documentation explaining what it does and what each parameter it needs means.

Usage

Micasense processing

Dataset/Flight structure

In order to perform a complete processing we need our dataset to comprise the following structure:

  • align_img:
    • Contains a capture that will be used to align the rest of the dataset capture (in tif format).
  • panel:
    • Contains the captures on the panel of the dataset for further calibration if necessary (in tif format).
  • raw_sky_imgs:
    • Contains the captures of the dataset panel for further calibration if necessary (in tif format).
  • raw_water_imgs:
    • Contains the captures of the flight in question (in tif format).
  • flight_lines.yaml:
    • Contains which captures define the usable lines for georeferencing, merging, etc...

Usage

See: mono_processing_jupyter/dji and batch_processing_jupyter/dji files

DJI processing

Dataset/Flight structure

In order to perform a complete processing we need our dataset to comprise the following structure:

  • bands:
    • Contains the captures of the flight in question (in tif format).
  • main:
    • Contains the captures of the flight in question (in jpg format).
  • summary.yml:
    • It is a file used to define flight metadata.
    • Its main function is to determine which captures will be georeferenced and joined.

Usage

See: mono_processing_jupyter/micasense and batch_processing_jupyter/micasense files