近期，隨著智慧型手機網上付款方面的生物特徵存取需求急劇增長。除了指靜脈識別系統外，無接觸式手部生物特徵系統，例如掌靜脈識別，已成為一種值得注意的方法。然而，無接觸式手部生物特徵系統面臨許多挑戰，包括不同的皮膚狀況、不規則的照明、手部厚度和色素等特徵變異、自然紋理的複雜性 、以及刺青、浮水印和陰影等造成的創傷或物理損害... 這些因素可能損害掌靜脈結構並導致識別錯誤。
在這個項目中，我們的目標是通過解決手部陰影問題來增強掌靜脈圖像。我們使用各種深度學習模型進行實驗，開發了一種手部陰影去除方法。該方法由三個部分組成：手部分割部分，用於從手部圖像中提取手部形狀；陰影檢測部分，用於生成二值陰影遮罩；以及陰影去除部分，從手部分割圖像、二值陰影遮罩和原始手部圖像中恢復無陰影的圖像。此外，我們不再使用未更改的陰影遮罩，而是提出一個隨機形狀生成器，以在每個訓練迭代中創建新的陰影遮罩和生成新鮮數據集。這種方法旨在豐富數據集並改善訓練過程。

There is a huge demand for biometric access in terms of online payments via smartphones recently. In addition to the finger vein recognition system, the contactless hand-based biometric system, such as palm vein recognition, has emerged as a notable approach. However, contactless hand-based biometric systems face numerous challenges, including different skin conditions, irregular illumination, variations in hand features such as thickness and pigmentation, complexity of natural texture, trauma or physical damages like tattoos, watermarks, and shadows... which can disrupt the palm vein architecture and lead to incorrect recognition.

In this project, our goal is to enhance palm vein images by addressing the issue of shadows on hand. We conducted experiments using various deep learning models to develop a method for shadow removal on the hand. This method consists of three parts: the hand segmentation part, which extracts the hand shape from the hand image; the shadow detection part, which is utilized to produced binary shadow mask; and the shadow removal part, which recover shadow-free image from hand segmentation image, binary shadow mask and original hand image. In addition, instead of using the unchanged shadow mask, we present a random shape generator to create a new shadow mask and generate a fresh dataset during each training iteration. This approach aims to enrich the dataset and improve the training process.

[1] D. Thapar, G. Jaswal, A. Nigam, and V. Kanhangad, "PVSNe: Palm Vein Authentication Siamese Network Trained using Triplet Loss and Adaptive Hard Mining by Learning Enforced Domain Spe­ cific Features," 2019 1st International Symposium on Bioelectronics and Biomedical Engineering (ISBBE), pp. 1-8, 2019. doi: 10.1109 / ISBA.2019 .8778623.
[2] D. Zhong, S. Liu, W. Wang, and X. Du, "Palm Vein Recognition with Deep Hashing Network: First Chinese Conference, PRCV 2018, Guangzhou, China, November 23-26, 2018, Proceedings, Part I," 2018. doi: 10.1007/978-3-030-03398-94.
[3] S. -J. Horng, D. -T. Vu, T. -V. Nguyen, W. Zhou, and C. -T. Lin, " Recognizing Palm Vein in Smartphones Using RGB Images," IEEE Transactions on Industrial Informatics, vol. 18, no. 9, pp. 5992-6002, Sept. 2022. doi: 10.1109 /TII.2021.3134016.
[4] J. Wang, X. Li, and J. Yang, "Stacked conditional generative adver­ sarial networks for jointly learning shadow detection and shadow re­ moval," Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018.
[5] H. Le and D. Samaras, "Shadow removal via shadow image decompo­ sition," in Proc. ICCV, 2019.
[6] H. Le and D. Samaras, "Physics-based shadow image decomposition for shadow removal," IEEE Trans. Pattern Anal. Mach. Intell. (2021).
[7] D. Dahmani, M. Cheref, and S. Larabi, "Zero-sum game theory model for segmenting skin regions," Image and Vision Computing, Volume 99, 2020, 103925, ISSN 0262-8856, https://doi. org/10.1016/ j.imavis.2020.103925.
[8] G. D. Finlayson, S. D. Hordley, C. Lu, and M. S. Drew, "On the re­ moval of shadows from images ," IEEE Transactions on Pattern Analy­ sis and Machine Intelligence (TPAMI), vol. 28, no. 1, pp. 59-68, 2005.
[9] G. D. Finlayson, M. S. Drew, and C. Lu, "Entropy minimization for shadow removal," International Journal of Computer Vision, vol. 85, no. l,pp.35-57,2009.
[10] R. Guo, Q. Dai, and D. Hoiem, "Paired regions for shadow detection and removal," IEEE Transactions on Pattern Analysis and Machine In­ telligence (TPAMI), vol. 35, no. 12, pp. 2956-2967, 2012.
[11] H. Gong and D. Cosker, "Interactive shadow removal and ground truth for variable scene categories," In Proceedings of British Machine Vision Conference (BMVC), 2014.
[12] M. Gryka, M. Terry, and G. J. Brostow, "Learning to remove soft shadows," ACM Transactions on Graphics (TOG), vol. 34, no. 5, 153, 2015.
[13] L. Qu, J. Tian, S. He, Y. Tang, and Rynson W.H. Lau, "DeshadowNet: A multi-context embedding deep network for shadow removal ," In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 4067-4075.
[14] Q. Zheng, X. Qiao, Y. Cao and R. W. H. Lau, "Distraction-Aware Shadow Detection," 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA, 2019, pp. 5162-5171, doi: 10.1109 /CVPR.2019.00531.
[15] J. Wu et al., "MedSegDiff: Medical Image Segmentation with Diffu­ sion Probabilistic Model," arXiv preprint arXiv:2211.00611, 2022.
[16] X. Hu, L. Zhu, C.-W. Fu, J. Qin, and P. A. Heng, "Direction­ aware spatial context features for shadow detection," In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018, pp. 7454-7462.
[17] S. Bell, C. Lawrence Zitnick, Kavita Bala, and Ross Girshick, " Inside-Outside Net: Detecting objects in context with skip pooling and recurrent neural networks ," In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 2874- 2883.
[18] I. Goodfellow et al., "Generative adversarial networks," Communi­ cations of the ACM, vol. 63, no. 11, pp. 139-144, 2020.
[19] X. Hu, Y. Jiang, C.-W. Fu, and P.-A. Heng, "Mask-ShadowGAN: Learning to remove shadows from unpaired data," In Proceedings of IEEE International Conference on Computer Vision (ICCV), 2019, pp. 2472-2481.
[20] B. Ding, C. Long, L. Zhang, and C. Xiao, "ARGAN: Attentive re­ current generative adversarial network for shadow detection and re­ moval," In Proceedings ofIEEE International Conference on Computer Vision (ICCV), 2019, pp. 10213-10222.
[21] L. Zhu et al., "Bidirectional feature pyramid network with recurrent attention residual modules for shadow detection," In Proceedings of the European Conference on Computer Vision (ECCV), 2018.
[22] L. Zhang et al., "Ris-gan: Explore residual and illumination with gen­ erative adversarial networks for shadow removal," In Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, no. 07, 2020.
[23] Y. He et al., "Unsupervised portrait shadow removal via generative priors," In Proceedings of the 29th ACM International Conference on Multimedia, 2021.
[24] J. Jing, Z. Wang, M. Ratsch, H. Zhang, "Mobile-Unet: An efficient convolutional neural network for fabric defect detection," Textile Re­ search Journal, vol. 92, no. 1-2, pp. 30-42, 2022.
[25] J. Ho, A. Jain, and P. Abbeel, "Denoising diffusion probabilistic mod­ els," Advances in Neural Information Processing Systems 33, 6840- 6851 (2020).
[26] Jun-Yan Zhu, Taesung Park, Phillip Isola, and Alexei A. Efros, " Unpaired image-to-image translation using cycleconsistent adversar­ ial networks," In ICCV, pages 2223-2232, 2017.
[27] Y.-Y. Chuang, D. B. Goldman, B. Curless, D. H. Salesin, and R. Szeliski, "Shadow matting and compositing," ACM Transactions on Graphics, vol. 22, no. 3, pp. 494-500, July 2003.
[28] T. Nagae, R. Abiko, T. Yamaguchi, and M. Ikehara, "Shadow Detec­ tion and Removal Using GAN," 2020 28th European Signal Processing Conference (EUSIPCO), Amsterdam, Netherlands, 2021, pp. 630-634, doi: 10.23919/Eusipco47968.2020.9287528.
[29] ISTD dataset, Available at: https://github.com/Deeplnsight-PCALab / ST-CGAN
[30] GTEA dataset, Available at: https://cbs.ic.gatech.edu/fpv /
[31] Idea of creating random shape, Available at: https:// stackover­ flow.com/ a/50751932/3237302
[32] llK Hands dataset, Available at: https:// sites.google.com/ view/ llkhands
[33] Bezier Curve, Available at: https:// en.wikipedia.org/ wiki/Bezier curve