A Hybrid Top-Hat and Frangi Filter Approach for Leaf Vein-Based Tree Species Identification Using Resnet-18

Accurate tree species identification is important for biodiversity conservation, forest management, and ecological monitoring, yet conventional identification based on visible leaf characteristics is time-consuming and dependent on expert knowledge. Leaf venation provides a structurally informative feature for automated identification, but fine veins can be difficult to extract consistently because of uneven illumination, noise, background interference, and image-quality variation. This study proposes a hybrid image-processing and deep-learning framework that combines the Top-Hat Transform and Frangi Filter for leaf-vein enhancement with a ResNet-18 convolutional neural network for species classification. A total of 1,190 leaf images representing five tree species Bauhinia racemosa (Apta), Ficus benghalensis (Vad), Ficus religiosa (Pimpal), Psidium guajava (Guajava), and Mangifera indica (Mangos) were obtained from public Kaggle and Mendeley datasets. The images were processed in MATLAB R2023a through leaf segmentation, grayscale conversion, Gaussian filtering, contrast enhancement, median filtering, Top-Hat transformation, Frangi filtering, image inversion, and region-of-interest extraction. The dataset was evaluated using an 80:20 stratified split, while an additional external evaluation used 50 online leaf images captured under uncontrolled conditions. Under the controlled dataset, the proposed framework achieved 95.87% classification accuracy, with validation accuracy reaching approximately 90.34% during training. The external 50 leaf evaluation achieved 54% accuracy, despite an average CNN confidence of 85.87%, revealing a substantial generalisation gap. Ficus religiosa achieved 10/10 correct classifications externally, whereas Mangifera indica and Psidium guajava were frequently misclassified as Ficus religiosa. The findings indicate that hybrid vein enhancement can support high-performing species classification on curated images, but broader and more diverse real-world training data are required before field deployment. The study therefore provides a practical foundation for semi-automated tree species identification while highlighting the need for stronger domain generalisation

leaf vein extraction, Top-Hat Transform, Frangi Filter, ResNet-18, tree species identification

1. Abdulazeez, A., et al. (2024). Leaf identification based on shape, color, texture and veins using probabilistic neural network. Computación y Sistemas, 28(3), 789–803. [Google Scholar] [Crossref]

2. Archana, S., & Jeevaraj, S. (2023). Deep learning-based plant species identification using leaf images. Ecological Informatics, 75, 101962. [Google Scholar] [Crossref]

3. Azlah, M. A. F., Chua, L. S., Rahmad, F. R., Abdullah, F. I., & Wan Alwi, S. R. (2019). Review on techniques for plant leaf classification and recognition. Computers, 8(4), 77. [Google Scholar] [Crossref]

4. Babbar, N., Jain, S., & Vikram, A. (2025). Advanced leaf vein pattern extraction and analysis using machine learning algorithms. International Journal of Research in Applied Science and Engineering Technology, 13(3), 68083. [Google Scholar] [Crossref]

5. Chouhan, S. S., Singh, U. P., Sharma, U., & Jain, S. (2019). A database of leaf images: Practice towards plant conservation with plant pathology. Mendeley Data, V1. [Google Scholar] [Crossref]

6. Cope, J. S., Corney, D., Clark, J. Y., Remagnino, P., & Wilkin, P. (2012). Plant species identification using digital morphometrics: A review. Expert Systems with Applications, 39(8), 7562–7573. [Google Scholar] [Crossref]

7. Frangi, A. F., Niessen, W. J., Vincken, K. L., & Viergever, M. A. (1998). Multiscale vessel enhancement filtering. In W. M. Wells et al. (Eds.), Medical Image Computing and Computer-Assisted Intervention—MICCAI'98 (pp. 130–137). Springer. [Google Scholar] [Crossref]

8. Gonzalez, R. C., & Woods, R. E. (2018). Digital Image Processing (4th ed.). Pearson. [Google Scholar] [Crossref]

9. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. [Google Scholar] [Crossref]

10. He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 770–778). [Google Scholar] [Crossref]

11. Grinblat, G. L., Uzal, L. C., Larese, M. G., & Granitto, P. M. (2016). Deep learning for plant identification using vein morphological patterns. Computers and Electronics in Agriculture, 127, 418–424. [Google Scholar] [Crossref]

12. Karnik, R., Patil, S., & Kulkarni, V. (2023). Leaf vein structure-based plant species classification using image processing techniques. Computers and Electronics in Agriculture, 203, 107452. [Google Scholar] [Crossref]

13. Kaur, P., & Kaur, A. (2019). Improved leaf vein extraction using morphological operations and Frangi filter. International Journal of Computer Applications, 178(7), 20–25. [Google Scholar] [Crossref]

14. Khan, A., Khan, S., Zia, T., & Farooq, A. (2017). Tree species identification using leaf vein patterns. International Journal of Advanced Computer Science and Applications, 8(4), 314–321. [Google Scholar] [Crossref]

15. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 25, 1097–1105. [Google Scholar] [Crossref]

16. Li, X., Chen, Y., & Huang, Z. (2023). Hybrid deep learning and image processing for leaf vein segmentation in natural environments. Computers and Electronics in Agriculture, 205, 107567. [Google Scholar] [Crossref]

17. Liao, Y.-H., & Zhang, S. (2024). Combining multispectral and high-resolution 3D imaging for leaf vein segmentation and density measurement. Frontiers in Plant Science. Manuscript submitted for publication. [Google Scholar] [Crossref]

18. Lin, H., Liu, X., Han, Z., Cui, H., & Dian, Y. (2023). Identification of tree species in forest communities at different altitudes based on multi-source aerial remote sensing data. Applied Sciences, 13(8), 4911. [Google Scholar] [Crossref]

19. Noh, Z. M., Saad, N. M., & Ramli, A. R. (2018). Leaf vein extraction techniques: A comparative study. Journal of Telecommunication, Electronic and Computer Engineering, 10(1–4), 121–126. [Google Scholar] [Crossref]

20. Sahoo, P. K., Soltani, S., & Wong, A. (1988). A survey of thresholding techniques. Computer Vision, Graphics, and Image Processing, 41(2), 233–260. [Google Scholar] [Crossref]

21. Satti, V., Satya, A., & Sharma, S. (2013). An automatic leaf recognition system for plant identification using machine vision technology. International Journal of Engineering Science and Technology, 5(4), 874. [Google Scholar] [Crossref]

22. Too, E. C., Yujian, L., Njuki, S., & Yingchun, L. (2019). A comparative study of fine-tuning deep learning models for plant disease identification. Computers and Electronics in Agriculture, 161, 272–279. [Google Scholar] [Crossref]

23. Wäldchen, J., & Mäder, P. (2018). Plant species identification using computer vision techniques: A systematic literature review. Archives of Computational Methods in Engineering, 25(2), 507–543. [Google Scholar] [Crossref]

24. Zheng, X., & Wang, X. (2010). Leaf vein extraction based on gray-scale morphology. International Journal of Image, Graphics and Signal Processing, 2(2), 25–31. [Google Scholar] [Crossref]

25. Zhu, X., et al. (2020). Automated extraction and measurement of leaf vein networks using object-oriented classification. Frontiers in Plant Science, 11, 499. [Google Scholar] [Crossref]

• LeafQuest: A Mobile-Based Augmented Reality for Plant Placement, Discovery, and Growth
• Participatory Ergonomic Intervention Approach on Musculoskeletal Disorder (MSD) in Construction Sectors: A Systematic Review
• Integrating GIS into Traffic Incident Management: A Web-Based System
• RideSmart: A Personalized Motorcycle Product Recommendation System Using TF-IDF and Descriptive Analytics for Javidson Motorshop
• Educational Technology Course Design in Pre-Service Teachers Education: A Bibliometric Review of the Research Landscape