reflectdetect 0.1.10

An automated workflow for in-flight radiometric calibration of UAV imagery

ReflectDetect

An automated workflow for in-flight radiometric calibration of UAV imagery

Report Bug · Request Feature

Documentation

Overview

Welcome to the official repository for the paper, "Application Note: An automated workflow for in-flight radiometric calibration of UAV imagery".

Abstract

UAVs equipped with optical sensors have transformed remote sensing in vegetation science by providing high-resolution, on-demand data, enhancing studies in forestry, agriculture, and environmental monitoring. However, accurate radiometric calibration of UAV imagery remains challenging due to environmental variability and the limitations of existing methods, such as the common single gray reference panel, which is prone to errors. ReflectDetect, an open-source tool, addresses these challenges by automating radiometric calibration using geotagging and AprilTag detection. This dual-module approach ensures reliable calibration under varying conditions, reduces human error, and increases efficiency through a user-friendly CLI. ReflectDetect's modular design supports future enhancements and broader applicability across research fields.

[!NOTE] We provide two workflows. For a detailed look at the technical details make sure to follow the upcoming setup and usage sections for your preferred workflow.

:artificial_satellite: Workflow 1: Geolocation-Based Calibration

1. Panel Reflectance Data: Gather reflectance values of the panels for the bands you will capture images in, either from the manufacturer (for commercial panels) or using a field spectrometer (for DIY panels). Save these values in a panel_properties.json file.
2. Field Setup: Position the calibration panels in the field.
3. Panel Location Data: Capture the exact locations of the panel corners (we used a Device Name in our testing) and save them in a panel_locations.gpkg file.
4. Image Capture: Fly your drone mission, capturing images.
5. Image Processing: Convert the captured images to orthophotos by rectifying and geo-referencing them (we used Software Name in our testing).
6. Run ReflectDetect: Use ReflectDetect to convert the orthophotos to reflectance data.

:white_square_button: Workflow 2: AprilTag-Based Calibration

1. Panel Reflectance Data: Gather reflectance values of the panels for the bands you will capture images in, either from the manufacturer (for commercial panels) or using a field spectrometer (for DIY panels). Save these values in a panel_properties.json file.
2. Print AprilTags: Print an AprilTag for each panel. PDF files for printing are available in the /apriltag_printouts/ directory of this repository.
3. Field Setup: Position the calibration panels and place the AprilTags according to the placement guide.
4. Image Capture: Fly your drone mission, capturing images.
5. Run ReflectDetect: Use ReflectDetect to convert the images to reflectance data.

Key Concepts and Vocabulary

• Panel: Calibration sheets placed in the field, used to compare captured intensity values in images with known reflectance values in the panel_properties.json file.

• AprilTag / Tag: A visual marker used for accurate detection in images.

• Field: The area captured by drone imagery.

• Image: Individual bands of captured data. ReflectDetect assumes all images were taken at equal time intervals and named in the format *_{band_index}.tif.
  For example, IMG_0052_6.tif indicates the 6th band.

• Orthophoto / Photo: An image where all bands are combined and geo-referenced.

Installation

Installing Python

To get started with this project, you'll need to have Python installed on your system. Follow the steps below to install Python:

1. Check if Python is Already Installed

Before installing, check if Python is already installed on your system:

• Windows/Mac/Linux:

  Open a terminal or command prompt and run the following command:

  python --version
  

  or

  python3 --version
  

  If Python is installed, you will see a version number. (Python 3.10 or higher is required)

2. Download and Install Python

If Python is not installed, follow these steps:

• Windows:

  1. Go to the official Python website.
  2. Download the latest version for Windows.
  3. Run the installer. Make sure to check the option "Add Python to PATH" during installation.
  4. Complete the installation process.

• Mac:

  1. Download the latest version of Python from the Python website.
  2. Open the downloaded .pkg file and follow the instructions to install.
  3. Alternatively, you can use Homebrew:

     brew install python
     

• Linux:

  1. Use the package manager for your distribution (e.g., apt for Ubuntu):

     sudo apt update
     sudo apt install python3
     

  2. For other distributions, refer to your package manager's documentation.

3. Verify Installation

After installation, verify that Python was installed correctly:

• Open a terminal or command prompt and run:

  python --version
  

  or

  python3 --version
  

  You should see the version of Python that you installed. Now you're ready to install project dependencies and start coding!

Installing ExifTool

ExifTool is an essential tool for working with image metadata in ReflectDetect. Follow the instructions below to install ExifTool on your system.

Windows Installation

1. Download ExifTool:

   • Visit the ExifTool Home Page and download the latest 64bit Windows Executable (exiftool-xx.xx_64.zip).

2. Extract the Zip File:

   • Double-click the downloaded .zip file to open it, then drag the exiftool-xx.xx_64x folder to your Desktop ( where xx represents the version).

3. Prepare for Command Line Use:

   • Open the exiftool-12.96_xx folder on your Desktop.
   • Rename the exiftool(-k).exe file to exiftool.exe.
   • Move the folder folder to a convenient location (e.g., C:\ExifTool\ or another folder of your choice).

4. Add the Folder to the PATH:

   • Open the Start Menu, search for "Environment Variables" and select Edit the system environment variables.
   • In the window that opens, click the Environment Variables button.
   • Under the User variables section, find the variable named Path and select it. Then click Edit.
   • In the new window, click New and enter the path to the folder where you moved exiftool.exe (e.g., C:\ExifTool).
   • Click OK to close all the windows.

5. Verify Installation:

   • Open Command Prompt (cmd) and type exiftool. You should see ExifTool's help information, confirming it is installed and recognized by your system.

macOS Installation

1. Download ExifTool:

   • Visit the ExifTool Home Page and download the latest MacOS Package (ExifTool-xx.xx.pkg).

2. Install ExifTool:

   • Double-click the downloaded .pkg file and follow the installation instructions.

3. Verify Installation:

   • Open Terminal and type exiftool. You should see ExifTool's help information, confirming it is installed and recognized by your system.

Linux Installation

1. Install via Package Manager:

   • On most Linux distributions, you can install ExifTool using the package manager. For example, on Ubuntu, run:

     sudo apt-get install libimage-exiftool-perl
     

2. Manual Install:

   • For manual installation instruction visit the ExifTool Install Page

3. Verify Installation:

   • Open your terminal and type exiftool. You should see ExifTool's help information, confirming it is installed and recognized by your system.

Installing reflectdetect

To install the reflectdetect CLI tools to your system, open a command line or terminal and run

pip install reflectdetect

Now reflectdetect-apriltag and reflectdetect-geolocation should be available as CLI tools.

:artificial_satellite: Workflow 1: Geolocation-Based Calibration

:artificial_satellite: Setup

Create a panel_properties.json file

To access the information about your calibration panels, we need you to create a panel_properties.json file. It includes the reflectance values of each panel for each of the bands you captured. In the following example we show how two panels might be configured. All the information about the first panel is between the first { } and so on.

The layer name corresponds to the name of the layer the coordinates of the panel corners are stored in, in the panel_locations.gpkg file

{
  "default_panel_width": 1.3,
  "default_panel_height": 1.3,
  "panel_properties": [
    {
      "layer_name": "corner_27",
      "bands": [
        0.9,
        1.0,
        1.0,
        1.0,
        0.72,
        0.91
      ]
    },
    {
      "layer_name": "corner_28",
      "bands": [
        0.9,
        1.0,
        0.43,
        0.70,
        0.35,
        0.4
      ]
    }
  ]
}

Create dataset folder

In order for reflect-detect to be able to gather the necessary information about the images, panels, camera, etc. , reflect-detect expects you to structure your data in the following format:

dataset_folder
│   panels_properties.json
│   panel_locations.gpkg
│
└───orthophotos
│   │   IMG_0000.tif
│   │   IMG_0001.tif
│   │   IMG_0002.tif
│   │   IMG_0003.tif
|   |   ...

:artificial_satellite: Usage

After preparing the dataset folder, you are ready to run reflectdetect. Open a command line or terminal.

To print the available arguments, run reflectdetect-geolocation --help

Assuming:

• the prepared dataset folder is at C:\Users\username\Desktop\dataset_folder
• the panel_properties.json and panel_locations.gpkg files and the images folder are in the dataset folder and correctly named

Minimal example:

To start the program, open your terminal or command line and run

cd C:\Users\username\Desktop\dataset_folder

then

reflectdetect-geolocation

Alternatively you can run the program from anywhere using

reflectdetect-geolocation "C:\Users\username\Desktop\dataset_folder"

:white_square_button: Workflow 2: AprilTag-Based Calibration

:white_square_button: Printing the Apriltags

:white_square_button: Placement Guide

:white_square_button: Measurement Guide

See: apriltag_area_measurement.ipynb

:white_square_button: Setup

Create a panel_properties.json file

To give access to the information about your calibration panels, create a panel_properties.json file. It includes the reflectance values of each panel for each of the bands you captured. The following examples show how two panels might be configured. All the information about the first panel is between the first { } and so on.

Minimal Example

{
  "default_panel_width": 1.3,
  "default_panel_height": 1.3,
  "panel_properties": [
    {
      "tag_id": 1,
      "bands": [
        0.9,
        1.0,
        1.0,
        1.0,
        0.72,
        0.91
      ]
    },
    {
      "tag_id": 2,
      "bands": [
        0.9,
        1.0,
        0.43,
        0.70,
        0.35,
        0.4
      ]
    }
  ]
}

Customizing the parameters

The first panel will use the default values, while the second panel specifies some of its own parameters. Only default_panel_width and default_panel_height are required, the other parameters will have the default values as below if not specified. The tag id has to correspond to the id of the apriltag you placed next to the given panel. No id can be used twice!

{
  "default_panel_width": 1.3,
  "default_panel_height": 1.3,
  "default_tag_family": "tag25h9",
  "default_tag_direction": "up",
  "default_panel_smudge_factor": 0.8,
  "default_tag_smudge_factor": 1.0,
  "default_shrink_factor": 0.8,
  "panel_properties": [
    {
      "tag_id": 1,
      "bands": [
        0.9,
        1.0,
        1.0,
        1.0,
        0.72,
        0.91
      ]
    },
    {
      "tag_id": 2,
      "panel_width": 2.0,
      "shrink_factor": 0.5,
      "bands": [
        0.9,
        1.0,
        0.43,
        0.70,
        0.35,
        0.4
      ]
    }
  ]
}

[!TIP] The default parameters can also be overwritten using CLI arguments

Run reflectdetect-apriltag -h to get a list

Create dataset folder

In order for reflect-detect to be able to gather the necessary information about the images, panels, etc. , reflect-detect expects you to structure your data in the following format:

dataset_folder
│   panels_properties.json
│   
└───images
│   │   IMG_0000_1.tif
│   │   IMG_0000_2.tif
│   │   IMG_0001_1.tif
│   │   IMG_0001_2.tif
|   |   ...

[!TIP] If any of the folders/files are located elsewhere, you can specify their location using the --panel_properties_file or --images_folder argument

:white_square_button: Usage

After preparing the dataset folder, you are ready to run reflectdetect. Open a command line or terminal.

To print the available arguments, run reflectdetect-apriltag --help

Assuming:

• the prepared dataset folder is at C:\Users\username\Desktop\dataset_folder
• the panel_properties.json file and the images folder are in the dataset folder and correctly named

Minimal example:

To start the program, open your terminal or command line and run

cd C:\Users\username\Desktop\dataset_folder

then

reflectdetect-apriltag

Alternatively you can run the program from anywhere using

reflectdetect-apriltag "C:\Users\username\Desktop\dataset_folder"

Planned Features

• Support for unequal time intervals between images
• Customize parameters on a per panel basis
• Add dataset verification script
• Remove 5% of outliers in panel intensities value

Contributing

Gallery

AI Usage Card

References

License