T. Chirakitsakul et al.: Integration of Convolutional Neural Network and Image Processing
V. CONCLUSION AND FUTURE WORKS This study presented a novel methodology for fibril detection and fibrillation index computation by integrating adaptive image processing techniques with deep learning to address the inherent challenges in analyzing microscopic fibrils in pulp fibers. Leveraging convolutional neural networks (CNNs) and patch-based analysis, the method effectively mitigates issues such as low contrast, noise, and fibril morphology variability, enabling accurate detection even under challenging imaging conditions. Experimental results demonstrate significant advancements over traditional man- ual and semi-automated techniques, and achieve a reliable correlation between the computed fibrillation index and expert-labeled ground truth. The average absolute deviation of 0.449% underscores its reliability across diverse refining conditions and image qualities. While the proposed approach provides significant improvements in detection accuracy and efficiency, some limitations remain, particularly in handling low-contrast fibrils and artifacts resembling fibrils. These challenges present opportunities for future enhancements, including improved feature extraction techniques, the expansion of training datasets, and the integration of advanced noise- handling algorithms. In conclusion, this study represents a significant step forward in automating fibril detection, offering a scalable, reliable, and efficient solution for industrial applications in pulp and paper manufacturing. Beyond its immediate application, the proposed methodology holds potential for broader use in materials science, biomedical research, and other domains requiring precise analysis of small, intricate structures. We hope that this research will be one of AI-driven fibril analysis studies that will lead to future innovations that require the analysis of fibril-like microstructures, advancing both quality control and scientific discovery. REFERENCES [1] M. Kauppinen, F. M. A. Varkaus, and E. Tiikkaja, ‘‘Increasing needs for fiber quality measurements–results from tests for new solutions in quality control,’’ Metso Autom. Tech. Rep. , pp. 1–9, 2006. [Online]. Available: https://www.eucalyptus.com.br/artigos/2006_Need_Fiber+Measurements. pdf [2] P. Przybysz, M. Dubowik, E. Małachowska, M. Kucner, M. Gajadhur, and K. Przybysz, ‘‘The effect of the refining intensity on the progress of internal fibrillation and shortening of cellulose fibers,’’ BioResources , vol. 15, no. 1, pp. 1482–1499, Jan. 2020. [3] A. Ushakov, Y. Alashkevich, V. Kozhukhov, and R. Marchenko, ‘‘Role of external fibrillation in high-consistency pulp refining,’’ BioResources , vol. 18, no. 3, pp. 5494–5511, Jul. 2023. [4] J. Viguié, S. Kumar, and B. Carré, ‘‘A comparative study of the effects of pulp fractionation, refining, and microfibrillated cellulose addition on tissue paper properties,’’ BioResources , vol. 17, no. 1, pp. 1507–1517, Jan. 2022. [5] S. R. Österling, ‘‘Distributions of fiber characteristics as a tool to evaluate mechanical pulps,’’ Ph.D thesis, Dept. Chem. Eng., Fac. Sci., Mid Sweden University, Sundsvall, Sweden, 2015. [6] T. Kang and H. Paulapuro, ‘‘Effect of external fibrillation on paper strength,’’ in Proc. 91st PAPTAC Annu. Meeting , vol. 107, Montreal, QC, Canada, Jan. 2005, pp. 51–54. [7] R. J. Kerekes, D. Mcdonald, and F. P. Meltzer, ‘‘External fibrillation of wood pulp,’’ TAPPI J. , vol. 22, no. 6, pp. 363–371, Jul. 2023.
[8] T. Kang, ‘‘Role external fibrillation pulp paper properties,’’ Helsinki University of Technology, Espoo, Finland, Tech. Rep., 2007. [9] I. Filipova, L. Andze, M. Skute, J. Zoldners, I. Irbe, and I. Dabolina, ‘‘Improving recycled paper materials through the incorporation of hemp, wood virgin cellulose fibers, and nanofibers,’’ Fibers , vol. 11, no. 12, p. 101, Nov. 2023. [10] F. Zambrano, H. Starkey, Y. Wang, C. Abbati de Assis, R. Venditti, L. Pal, H. Jameel, M. Hubbe, O. Rojas, and R. Gonzalez, ‘‘Using micro- and nanofibrillated cellulose as a means to reduce weight of paper products: A review,’’ BioResources , vol. 15, no. 2, pp. 4553–4590, May 2020. [11] S. H. Osong, S. Norgren, P. Engstrand, M. Lundberg, M. Reza, and V. Tapani, ‘‘Qualitative evaluation of microfibrillated cellulose using the crill method and some aspects of microscopy,’’ Cellulose , vol. 23, no. 6, pp. 3611–3624, Dec. 2016. [12] Techpap. (2024). Fiber and Shive Analyzer Morfi Neo . Accessed: Sep. 27, 2024. [Online]. Available: https://www.techpap.com/fiber-and- shive-analyzer-morfi-neo,lab-device,31.html [13] U. Rakesh and N. Farrand, ‘‘New developments in online pulp quality measurements,’’ Indian Pulp Paper Tech. Assoc. (IPPTA) , vol. 25, no. 1, pp. Jan.–Mar., 2013. [14] H. R. Motamedian, A. E. Halilovic, and A. Kulachenko, ‘‘Mechanisms of strength and stiffness improvement of paper after PFI refining with a focus on the effect of fines,’’ Cellulose , vol. 26, no. 6, pp. 4099–4124, Apr. 2019. [15] R. Sunkara and T. Luo, ‘‘No more strided convolutions or pooling: A new CNN building block for low-resolution images and small objects,’’ in Machine Learning and Knowledge Discovery in Databases , M.-R. Amini, S. Canu, A. Fischer, T. Guns, P. Kralj Novak, and G. Tsoumakas, Eds., Cham, Switzerland: Springer, 2023, pp. 443–459. [16] X. He, X. Zheng, X. Hao, H. Jin, X. Zhou, and L. Shao, ‘‘Improving small object detection via context-aware and feature-enhanced plug-and- play modules,’’ J. Real-Time Image Process. , vol. 21, no. 2, pp. 1–12, Apr. 2024. [17] K. Hyll, ‘‘Image-based quantitative infrared analysis and microparticle characterisation for pulp and paper applications,’’ Ph.D thesis, Dept. Prod. Eng., Metrology Optics, School Ind. Eng. Manage. (ITM), KTH Royal Inst. Technol., Stockholm, Sweden, 2016. [18] Optest. (2024). Fiber Quality Analyzer 360 (fqa-360) . Accessed: Sep. 27, 2024. [Online]. Available: https://optest.com/pulp-and-fibrous- materials/fiber-quality-analyzer-360-fqa-360 [19] L&w Fiber Tester , ABB Inc., Zurich, Switzerland, 2016. [20] G. Eymin-Petot-Tourtollet, F. Cottin, A. Cochaux, and M. Petit-Conil, ‘‘The use of morfi analyser to characterise mechanical pulps, international mechanical pulping conference,’’ in Proc. Poster Presentations , 2003, pp. 225–232. [21] J. Pennells, B. Heuberger, C. Chaléat, and D. J. Martin, ‘‘Assessing cellulose micro/nanofibre morphology using a high throughput fibre analysis device to predict nanopaper performance,’’ Cellulose , vol. 29, no. 4, pp. 2599–2616, Mar. 2022. [22] S. Yacoub and A. Haapala. (Apr. 2020). Digging More Deeply Into Fibers and Particles . [Online]. Available: https://www.valmet.com/insights/ articles/automation/digging-more-deeply-into-fibers-and-particles/ [23] B. Hou, P. Li, and Z. Song, ‘‘Edge detection of pulp fibre image based on directional wavelet transform,’’ in Proc. 5th World Congr. Intell. Control Autom. , Jun. 2004, pp. 4034–4036. [24] B.-P. Hou and W. Zhu, ‘‘The application of directional wavelets in multiscale representation of pulp fibre image,’’ in Proc. Int. Conf. Mach. Learn. Cybern. , vol. 7, Sep. 2004, pp. 4314–4318. [25] S. Qiu and J. Bian, ‘‘Curvelet transform and its application in the analysis of pulp fibre images,’’ in Proc. IEEE Int. Conf. Inf. Acquisition , Apr. 2006, pp. 672–676. [26] J. Bian and S. Qiu, ‘‘Pulp fibre recognition based on curvelet transform and skeleton tracing algorithm,’’ in Proc. 2nd IEEE Conf. Ind. Electron. Appl. , May 2007, pp. 2534–2538. [27] L. Qin and J. Jian, ‘‘The recognition of linear objects based on edge detection and skeleton tracking,’’ in Proc. 3rd Int. Symp. Inf. Sci. Eng. , Dec. 2010, pp. 379–381. [28] H. Huttunen, H. Oinonen, J. Selinummi, P. Ruusuvuori, and V. Voipio, ‘‘Polynomial pulp fiber modeling,’’ in Proc. 20th Eur. Signal Process. Conf. (EUSIPCO) , Aug. 2012, pp. 2099–2103. [29] V. Parvathy, S. Shankar, L. Thomas, and C. Arun, ‘‘Characterization of paper pulp fiber using image processing techniques,’’ Int. J. Eng. Res. Technol. , vol. 3, no. 4, pp. 1–6, 2014.
74645
VOLUME 13, 2025
Made with FlippingBook. PDF to flipbook with ease