This repository contains the official implementation of the AWRP encoding process from the paper:

"Enhancing EEG-Based Emotion Recognition Using Asymmetric Windowing Recurrence Plots"
Published in IEEE Access, 2024
DOI: 10.1109/ACCESS.2024.3409384

• AWRP_encoding_DEAP.py: AWRP generator for the DEAP dataset (32 subjects, 40 trials each, 32 channels).
• AWRP_encoding_SEED_sub1.py: AWRP generator for subject 1 of the SEED dataset (15 trials, 64 channels).

This implementation converts EEG signals into Asymmetric Windowing Recurrence Plots (AWRPs) by:

1. Segmenting EEG signals into fixed-length windows.
2. Generating recurrence plots (RPs) from each segment.
3. Combining adjacent RPs into averaged RPs (ARPs).
4. Arranging ARPs into a 2D image grid to form the final AWRP.
5. Saving AWRPs as .png images per subject-trial-channel.

• Python 3.8+
• NumPy
• SciPy
• Matplotlib
• OpenCV (cv2)
• pyts
• Pillow
• psutil

Install with:

pip install numpy scipy matplotlib opencv-python pyts pillow psutil

For DEAP dataset (32 subjects):

python AWRP_encoding_DEAP.py

For SEED dataset (subject 1 only):

python AWRP_encoding_SEED_sub1.py

🔧 Note: The SEED script is tailored for subject 1. To process other subjects, trial key names (djc_eegX) must be updated accordingly based on the SEED dataset's structure.

• AWRPs are saved as .png images in folders:
  • AWRP_DEAP_8/
  • AWRP_SEED_8/
• Timing and memory usage logs are written to:
  • avg_time-memory_usage_AWRP_8.txt

If you use this code, please cite our paper:

@article{prabowo2024,
  author    = {Wahyu Prabowo, Dwi and Akhmad Setiawan, Noor and Debayle, Johan and Nugroho, Hanung Adi},
  journal   = {IEEE Access},
  title     = {Enhancing EEG-Based Emotion Recognition Using Asymmetric Windowing Recurrence Plots},
  year      = {2024},
  volume    = {12},
  doi       = {10.1109/ACCESS.2024.3409384}
}

This code is for academic use only. For other purposes, please contact the authors.