Génération de bases de données images IR sous contraintes avec variabilité thermique intrinsèque des cibles

Jerome Gilles,Stephane Landeau,Tristan Dagobert,Philippe Chevalier,Christian Bolut

2024-11-12

Abstract:In this communication, we propose a method which permits to simulate images of targets in infrared imagery by superimposition of vehicle signatures in background, eventually with occultants. We develop a principle which authorizes us to generate different thermal configurations of target signatures. This method enables us to easily generate huge datasets for ATR algorithms performance evaluation.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is: how to generate an infrared image database with the target of inherent thermal variability in order to evaluate and optimize the detection, recognition and classification (DRI) algorithms in infrared imaging.

Specifically, the author proposes a method to simulate infrared images by embedding vehicle signatures into background images and considering the inherent thermal variability of the target. This method aims to generate large - scale and diverse data sets for evaluating the performance of DRI algorithms. The following are the key points of this method:

1. **Generate realistic infrared images**:
- Generate realistic infrared images by embedding vehicle signatures in background images and adjusting the thermal configuration according to the operation scenario.
- Use actually photographed vehicle images, obtained under extreme conditions (such as ambient temperature and maximum operating temperature), and map them onto 3D models.

2. **Consider inherent thermal variability**:
- Propose a method to generate vehicle signatures in different thermal states by interpolating signatures at ambient temperature and operating temperature.
- The thermal state is controlled by the parameter \(\lambda\), which is defined as:
\[
T_{IR}=(1 - \lambda_R)T_{AR}+\lambda_TF_R
\]
where \(\lambda\in[0, 1]\) represents the level of thermal variability.

3. **Generate diverse target signatures**:
- Divide the vehicle into multiple sub - regions with consistent thermal behavior (such as the engine, the vehicle body, the exhaust pipe, etc.), and select different \(\lambda\) values to generate intermediate thermal states.
- Randomly select \(\lambda\) values within different intervals using Gaussian distribution or semi - Gaussian distribution to ensure the diversity of thermal states.

4. **Construct a large - scale data set**:
- Construct a data set containing a large number of target signatures by generating vehicle images from different angles.
- These data sets can be used for training and evaluating DRI algorithms and improving their performance under various operating conditions.

5. **Application and verification**:
- Verify the authenticity of the generated images and the effectiveness of the method through experiments.
- Show vehicle images with different thermal configurations in the same scenario and the comparison between real - collected images and simulated images, proving the practicality of the method.

In summary, this paper aims to improve the generation method of infrared image databases by introducing the inherent thermal variability of the target, thereby enhancing the evaluation and optimization effects of DRI algorithms.