SFP: A Spectral Filter Predictor with Progressive Feedback Method for Latent Fingerprint Restoration

We're sharing the executable file to ensure a user-friendly experience. Our software combines MATLAB and Python.

The software comprises two parts:

1. Preprocessing with Total Variation This step enhances latent fingerprints by reducing noise.

2. Progressive Feedback Method for Restoration This method combines a new spectral filter predictor within a feedback framework. The spectral filter predictor PyTorch model is converted to ONNX for CPU, no GPU required.

We understand the importance of transparency and reproducibility in research. However, variations in environments can cause differences in results. Therefore, we provide the results of NIST SD27, NIST SD302, IIITD MOLF, and IIITD MSLFD on https://skconan.github.io/SFP-Progressive-Feedback-Latent-Fingerprint-Restoration.

Requirements

• Windows 10 or 11 operating system.

• Storage 14 GB

  • ksip_lfp_enh_installer 300 MB
  • MATLAB_Runtime_R2022a_Update_6_win64 (installer 4 GB and install space required 8 GB)

Installation

Install MATLAB Runtime version R2022a (9.12)

1. Download MATLAB Runtime from www.mathworks.com Or MATLAB_Runtime_R2022a_Update_6_win64.zip.

2. Extract files and install MATLAB Runtime using setup.exe.

Install KSIP Latent Fingerprint Enhancement

1. Download ksip_lfp_enh_installer.exe.

2. Install KSIP LFP ENHANCEMENT using ksip_lfp_enh_installer.exe. The installation directory will be C:\Program Files (x86)\KSIP LFP ENHANCEMENT

3. Setup environment path

   • Go to Environment Variables
   • Add C:\Program Files (x86)\KSIP LFP ENHANCEMENT in the Path variable under System variables.

   If KSIP LFP ENHANCEMENT installed in a different location, add that specific path to System variables instead of C:\Program Files (x86)\KSIP LFP ENHANCEMENT.

Usage

0. Please ensure the latent fingerprint image resolution is set to 500 DPI.

1. Open Terminal or Windows Powershell

2. Run ksip_sfp.exe --help and Enter to show the program usage

    usage: ksip_sfp.exe [-h] [-s START_INDEX] [-e END_INDEX] -fp
                            FINGERPRINT_DIR [-seg SEGMENT_DIR] -out OUTPUT_DIR
   
    optional arguments:
    -h, --help            show this help message and exit
    -s START_INDEX, --start_index START_INDEX
                            Start index
    -e END_INDEX, --end_index END_INDEX
                            End index
    -fp FINGERPRINT_DIR, --fingerprint_dir FINGERPRINT_DIR
                            Fingerprint Directory
    -seg SEGMENT_DIR, --segment_dir SEGMENT_DIR
                            Segment Directory
    -out OUTPUT_DIR, --output_dir OUTPUT_DIR
                            Output Directory
   

Example

• Run NIST SD27 Enhancement

    ksip_sfp.exe -fp D:\NIST_SD27\Latent -seg D:\NIST_SD27\GlobalDict -out D:\NIST_SD27\Enhancement
  

• Run NIST SD27 Enhancement without Segment

    ksip_sfp.exe -fp D:\NIST\NIST_SD27\Latent -out D:\NIST_SD27\Enhancement
  

• Run NIST SD27 Enhancement without Segment first 10 images

    ksip_sfp.exe -fp D:\NIST_SD27\Latent -out D:\NIST_SD27\Enhancement -s 0 -e 10
  

• Run NIST SD27 Enhancement without Segment 10 images start at image no. 15

    ksip_sfp.exe -fp D:\NIST_SD27\Latent -out D:\NIST_SD27\Enhancement -s 15 -e 25
  

Output Example

0001/0001 | Image Path: C:\Users\Administrator\Desktop\test\img\001L2U.bmp
0001/0001 | Segment Path: C:\Users\Administrator\Desktop\test\seg\001L2U.png
0001/0001 | Start Enhancement
0001/0001 | Enhancement Success
0001/0001 | Save enhanced image to C:\Users\Administrator\Desktop\test\out\enhanced\001L2U.png
0001/0001 | Execution time: 26.39 second
0001/0001 | Average Execution time: 26.39 sec. Total time: 0:00:26.391823. Estimating Time to Complete: 2023-08-07 07:34:30.650199

Acknowledgements

This work was supported in part by the Department of Electrical Engineering, Faculty of Engineering, Kasetsart University, and in part by the Siew-Sngiem Karnchanachari Research Leadership and Young Professorship Awards.

License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Citing SFP

If you are using SFP or benchmarks in your research, kindly reference DOI: 10.1109/ACCESS.2024.3397729 the following.

@ARTICLE{10526230,
  author={Kriangkhajorn, Supakit and Horapong, Kittipol and Areekul, Vutipong},
  journal={IEEE Access}, 
  title={Spectral Filter Predictor for Progressive Latent Fingerprint Restoration}, 
  year={2024},
  volume={12},
  number={},
  pages={66773-66800},
  keywords={Fingerprint recognition;Image restoration;Friction;Frequency-domain analysis;Filtering;Image matching;Deep learning;Image restoration;Image forensics;Machine learning;Fingerprint recognition;image restoration;image enhancement;image filtering;image forensics;machine learning},
  doi={10.1109/ACCESS.2024.3397729}}

or

S. Kriangkhajorn, K. Horapong and V. Areekul, "Spectral Filter Predictor for Progressive Latent Fingerprint Restoration," in IEEE Access, vol. 12, pp. 66773-66800, 2024, doi: 10.1109/ACCESS.2024.3397729.

Contact

If you have any questions or need assistance, reach us [email protected] / [email protected].