DENOISING DIFFUSION PROBABILISTIC MODEL WITH WAVELET PACKET TRANSFORM FOR FINGERPRINT GENERATION

(Received: 29-Jun.-2024, Revised: 9-Aug.-2024 , Accepted: 12-Aug.-2024)

Authors Li Chen, Huah Yong Chan,

Keywords #Diffusion model #Fingerprint #Image processing #Wavelet packet transform

Abstract The majority of contemporary fingerprint synthesis is based on the Generative Adversarial Network (GAN). Recently, the Denoising Diffusion Probabilistic Model (DDPM) has been demonstrated to be more effective than GAN in numerous scenarios, particularly in terms of diversity and fidelity. This research develops a model based on the enhanced DDPM for fingerprint generation. Specifically, the image is decomposed into sub-images of varying frequency sub-bands through the use of a wavelet packet transform (WPT). This method enables DDPM to operate at a more local and detailed level, thereby accurately obtaining the characteristics of the data. Furthermore, a polynomial noise schedule has been designed to replace the linear noise strategy, which can result in a smoother noise-addition process. Experiments based on multiple metrics on the datasets SOCOFing and NIST4 demonstrate that the proposed model is superior to existing models.

References

[1] J. Priesnitz et al., "An Overview of Touchless 2D Fingerprint Recognition," EURASIP Journal on Imageand Video Processing, vol. 2021, no. 1, p. 8, Dec. 2021.

[2] J. Kolberg et al., "COLFISPOOF: A New Database for Contactless Fingerprint Presentation AttackDetection Research," Proc. of the 2023 IEEE/CVF Winter Conf. on Applications of Computer Vision Workshops (WACVW), pp. 653–661, Waikoloa, HI, USA, 2023.

[3] E. Prabakaran and K. Pillay, "Nanomaterials for Latent Fingerprint Detection: A Review," Journal ofMaterials Research and Technology, vol. 12, pp. 1856–1885, May 2021.

[4] M. Kücken M and A. Newell, "Fingerprint Formation," Journal of Theoretical Biology, vol. 235, no. 1,pp. 71–83, Jul. 2005.

[5] R. Cappelli et al., "Synthetic Fingerprint-image Generation," Proc. of the 15th Int. Conf. on PatternRecognition (ICPR-2000), vol. 3, pp. 471–474, Barcelona, Spain, 2000.

[6] J. L. Araque et al., "Synthesis of Fingerprint Images," Proc. of the 2002 Int. Conf. on Pattern Recognition,vol. 2, pp. 422–425, DOI: 10.1109/ICPR.2002.1048329, Quebec City, QC, Canada, 2002.

[7] J. Feng and A. K. Jain, "Fingerprint Reconstruction: from Minutiae to Phase," IEEE Transactions onPattern Analysis and Machine Intelligence, vol. 33, no. 2, pp. 209–223, 2011.

[8] T. Uz et al., "Minutiae-based Template Synthesis and Matching for Fingerprint Authentication,"Computer Vision and Image Understanding, vol. 113, no. 9, pp. 979–992, Sep. 2009.

[9] M. Attia, "Fingerprint Synthesis via Latent Space Representation," Proc. of the 2019 IEEE Int. Conf. onSystems, Man and Cybernetics (SMC), pp. 1855–1861, Bari, Italy, 2019.

[10] O. Striuk and Y. Kondratenko, "Adaptive Deep Convolutional GAN for Fingerprint Sample Synthesis,"Proc. of the 2021 IEEE 4th Int. Conf. on Advanced Information and Communication Technologies (AICT), pp. 193–196, DOI: 10.1109/AICT52120.2021.9628978, Lviv, Ukraine, 2021.

[11] P. Schuch et al., "De-convolutional Auto-encoder for Enhancement of Fingerprint Samples," Proc. of the2016 6th Int. Conf. on Image Processing Theory, Tools and Applications (IPTA), pp. 1–7, Oulu, Finland, DOI: 10.1109/IPTA.2016.7821036, 2016.

[12] S. Minaee and A. Abdolrashidi, "Finger-GAN: Generating Realistic Fingerprint Images UsingConnectivity Imposed GAN," arXiv, [Online], Available http://arxiv.org/abs/1812.10482, Dec. 2018.

[13] S. Seidlitz et al., "Generation of Privacy-friendly Datasets of Latent Fingerprint Images Using GenerativeAdversarial Networks," Proc. of the 16th Int. Joint Conf. on Computer Vision, Imaging and Computer Graphics Theory and Applications, pp. 345–352, DOI: 10.5220/0010251603450352, 2021.

[14] J. Ho et al., "Denoising Diffusion Probabilistic Models," Proc. of the 34th Int. Conf. on Neural InformationProcessing Systems, pp.6840-6851, Vancouver, BC, Canada, 2020.

[15] A. Ramesh et al., "Hierarchical Text-conditional Image Generation with CLIP Latents," arXiv, [Online],Available http://arxiv.org/abs/2204.06125, Apr. 2022.

[16] R. Rombach et al., "High-resolution Image Synthesis with Latent Diffusion Models," Proc. of the 2022IEEE/CVF Conf. on Computer Vision and Pattern Recognition (CVPR), pp. 10674–10685, New Orleans, LA, USA, 2022.

[17] Y. I. Shehu et al., "Sokoto Coventry Fingerprint Dataset," arXiv, [Online], Availablehttps://arxiv.org/abs/1807.10609, 2018.

[18] C.Watson, NIST Special Database 4, NIST 8-Bit Gray Scale Images of Fingerprint Image Groups, WorldWide Web-Internet and Web Information Systems, [Online], Available: https://www.nist.gov/publications/nist-special-database-4-nist-8-bit-gray-scale-images-fingerprint-image-groups, 2008.

[19] J. Sohl-Dickstein et al., "Deep Unsupervised Learning Using Nonequilibrium Thermodynamics," Proc.of the 32nd Int. Conf. on Machine Learning, vol. 37, pp.2256 - 2265, 2015.

[20] A. Nichol and P. Dhariwal, "Improved Denoising Diffusion Probabilistic Models," Proc. of the 38th Int.Conf. on Machine Learning, vol 139, pp.8162-8171, 2021.

[21] D. P. Kingma et al., "Variational Diffusion Models," Proc. of the 35th Int. Conf. on Neural InformationProcessing Systems, pp. 21696-21707, DOI: 10.5555/3540261.3541921, 2021.

[22] R. San-Roman et al., "Noise Estimation for Generative Diffusion Models," arXiv, [Online], Availablehttp://arxiv.org/abs/2104.02600, Sep. 2021.

[23] J. Wang et al., "Guided Diffusion Model for Adversarial Purification," arXiv, [Online], Availablehttp://arxiv.org/abs/2205.14969, Jun. 2022.

[24] J. Song et al., "Denoising Diffusion Implicit Models," Proc. of the 2021 Int. Conf. on LearningRepresentations, Vienna, Austria, [Online], Available https://openreview.net/forum?id=St1giarCHLP, 2021.

[25] P. Friedrich et al., "WDM: 3D Wavelet Diffusion Models for High-Resolution Medical Image Synthesis,"arXiv, [Online], Available http://arxiv.org/abs/2402.19043, Feb. 2024.

[26] Y. Huang et al., "WaveDM: Wavelet-based Diffusion Models for Image Restoration," IEEE Transactionson Multimedia, vol. 26, pp. 7058-7073, DOI: 10.1109/TMM.2024.3359769, 2024.

[27] K. Xu et al., "Stage-by-stage Wavelet Optimization Refinement Diffusion Model for Sparse-View CTReconstruction," IEEE Trans. Med. Imaging, DOI: 10.1109/TMI.2024.3355455, 2024.

[28] H. Phung et al., "Wavelet Diffusion Models Are Fast and Scalable Image Generators," Proc. of the 2023IEEE/CVF Conf. on Computer Vision and Pattern Recognition (CVPR), pp. 10199–10208, Vancouver, BC, Canada, 2023.

[29] N. B. Bynagari, "GANs Trained by a Two Time-Scale Update Rule Converge to a Local NashEquilibrium," Asian Journal of Applied Sciences & Eng., vol. 8, no. 1, pp. 25–34, Apr. 2019.

[30] S. Barratt and R. Sharma, "A Note on the Inception Score," Proc. ICML 2018 Workshop on TheoreticalFoundations and Applications of Deep Generative Models, Stockholm, Sweden, [Online], Available https://drive.google.com/file/d/1y--gMfGtjlNcQXTPm-vLhuqlUbh19nF9/view, 2018.