IRMapper is a Pyside6 application for creating thermal pointclouds from an RGB/IR image folder, using photogrammetry*. For users who are only interested in image processing, it also offers to process batch of thermal pictures with consistent parameters. If you are looking for pure thermal image processing (no 3D reconstruction), have a look here : https://github.com/s-du/Thermogram
🌡️📈
Note: At the moment, the application is compatible with the following drones models:
*The application uses Agisoft Metashape Python API but we are working on Open Drone Map integration!
**We are still looking for photosets taken with the DJI M3T or H20T to improve calibration!
The project is still in pre-release, so do not hesitate to send your recommendations or the bugs you encountered!
GUI for thermal photogrammetry
Point clouds with integrated infrared data can be generated for the exterior of the buildings, but also for the interior. In addition to highlighting potential thermal weaknesses of the envelope, the processing of such point clouds also allows new types of analysis.
Here, starting from a DJI image folder (typically ..._T.JPEG alternating with ..._W.JPEG), the application will guide the user to reconstruct a infrared point cloud
First, the collected IR images have to be modified before entering the photogrammetry process. This phase was essential because of the highly processed nature of the R-JPG images coming out of the DJI M2EA/ DJI M3T drone. The temperature scale is the most critical parameter here. In order to guarantee an optimal integration of the thermal information into the 3D process, it is imperative to have a fixed temperature scale on the whole photoset (which is not the case by default). This app allows batch processing of infrared images, making photosets more easily compatible with photogrammetric processes.
From infrared images with a consistant temperature range, it is possibile to obtain accurate 3D models. First, this implies that pairs of images are captured: Colour + Infrared images at each shot. This is the default behaviour of DJI thermal drones. Secondly, the photogrammetric process requires to have a significant overlap between individual shots. When regular RGB pictures are used, 75% overlap can be judged satisfactory. With IR/RGB pairs, it is advised to increase the overlap, because the field of view is smaller on IR pictures.
Difference in terms of field of view between thermal and colour pictures
The app offers the following key features:
Image Loading:
Image Visualization:
Image Enhancement:
Image batch processing:
3D reconstruction
3D Visualization
Upcoming key features:
Image Analysis:
3D reconstruction:
A simple Windows setup file is available in the 'Releases' section!
Alternatively:
Clone the repository:
git clone https://github.com/s-du/IRMapper
Navigate to the app directory:
cd IRMapper
(Optional) Install and activate a virtual environment
Install the required dependencies:
pip install -r requirements.txt
Run the app:
python main.py
(coming soon)
Contributions to the IRMapper App are welcome! If you find any bugs, have suggestions for new features, or would like to contribute enhancements, please follow these steps:
This project was made possible thanks to subsidies from the Brussels Capital Region, via Innoviris. Feel free to use or modify the code, in which case you can cite Buildwise and the Pointify project!