Advanced Leaf Vein Pattern Extraction and Analysis using Machine Learning Algorithms

Abstract

Leaf vein extraction has a significant utility in botanical research and plant taxonomy. This research proposes an automated approach for extracting and analysing vein patterns from leaf photos using image processing and machine learning approaches. The methodology involves photo capture, pre-processing, morphological procedures, and feature extraction with Local Binary Pattern (LBP). Various machine learning classifiers like as Logistic Regression, SVM, Decision Tree, Random Forest, KNN, Gaussian Naïve Bayes, Gradient Boosting, and XGBoost are applied for vein classification. Among these, Gradient Boosting and XGBoost obtain the maximum accuracy of 92.86%, followed by Random Forest with 90.48% accuracy. The suggested method offers high accuracy and efficiency, making it a great tool for plant species identification, disease detection, and ecological investigations.

Introduction

. Introduction

Leaf vein extraction is vital for botanical research, plant taxonomy, and ecological analysis. Veins provide structural support and help transport water and nutrients. Analyzing venation offers insights into plant health, physiology, and evolution.

Traditional methods for extracting leaf veins are manual or semi-automated, making them time-consuming, error-prone, and unsuitable for large-scale analysis. To overcome these challenges, this study proposes an automated system using image processing, morphological operations, and machine learning (ML).

2. Literature Review

The field has evolved from basic edge detection to advanced ML-based models:

• Fu and Chi: Two-stage fluorescent edge detection with neural network refinement.

• Li and Chi: Used Independent Component Analysis (ICA) for partial venation.

• Kirchgeßner & Jeyalakshmi: Applied B-splines and Canny edge detection respectively, but required manual input or were noise-sensitive.

• Craviotto & Mullen: Used hit-or-miss transformation and artificial ant algorithms.

• Mark Fricker: Applied CNNs for high-precision vein analysis.

• Feng, Clarke, Herdiyeni, and Hong: Introduced techniques using neural networks, scale-space filtering, Hessian matrices, and achieved high vein classification accuracy (up to 97.1%).

3. Proposed Methodology

A. Image Acquisition

• 277 high-resolution leaf images (144 Mango, 133 Jatropha).

• Dataset split: 85% training, 15% testing.

B. Pre-processing

• Convert to grayscale and enhance contrast using techniques like gamma correction, histogram equalization, and decorrelation stretching.

C. Image Segmentation

• Thresholding, region growing, and Sobel/Canny edge detection used to isolate veins.

D. Morphological Operations

• Erosion, dilation, and subtraction clean up the image and highlight vein structures.

E. Feature Extraction

• Local Binary Pattern (LBP) captures vein texture by analyzing pixel intensity relationships.

F. Machine Learning

• Eight models trained to classify vein structures based on extracted features.

G. Evaluation Metrics

• Models assessed using Accuracy, Precision, Recall, and F1-score.

4. Results and Analysis

Model Performance:

• Gradient Boosting & XGBoost: 92.86% accuracy — top-performing models.

• Random Forest: 90.48% — robust and effective.

• Decision Tree: 88.10% — good baseline.

• Gaussian Naïve Bayes & KNN: 69.05%

• SVM: 62%

• Logistic Regression: 61.90% — least effective.

Insights:

• Ensemble models (Random Forest, Gradient Boosting, XGBoost) significantly outperform basic classifiers due to better handling of complex patterns.

• High-performing models show balanced precision and recall, making them reliable for vein detection tasks.

• Simpler models like Logistic Regression and SVM struggle with the intricate nature of venation patterns.

Conclusion

The leaf vein extraction technique proposed in this research provides an automated and accurate approach for recognising and classifying vein patterns from leaf photographs. By merging image processing techniques, morphological operations, feature extraction using Local Binary Pattern (LBP), and machine learning algorithms, the system effectively isolates and analyzes vein patterns.The results reveal that ensemble and boosting algorithms outperform simpler models in vein classification. Both Gradient Boosting and XGBoost Classifiers attain the highest accuracy of 92.86%, making them the most dependable models for this task. Random Forest, with an accuracy of 90.48%, likewise displays strong performance, illustrating the usefulness of ensemble approaches.The system’s scalability and efficiency make it highly suitable for botanical research, plant taxonomy, and ecological investigations. Its capacity to process big datasets with minimum operator involvement decreases the time and effort necessary for vein extraction and classification. Furthermore, the automated approach provides consistent and reliable results, making it a significant tool for plant species identification and disease diagnosis. In summary, the suggested methodology provides a stable and scalable solution for leaf vein extraction, presenting substantial promise for future applications in agriculture, botany, and environmental monitoring

References

[1] Fu H., and Chi Z., “A two-stage approach for leaf vein extraction,” In: Proceedings of International Conference on Neural Networks and Signal Processing, vol. I, Nanjing, Jiangsu, China, December 12–15, 2003, pp. 208–211 [2] Y. Li, Z. Chi, D.D. Feng Leaf vein extraction using independent component analy-sis IEEE International Conference on Systems Man and Cybernetics (2006), pp. 3890- 3894 [3] Kirchgessner N., Scharr H., and Schurr U., “Robust vein extraction on plant leaf images,” In: 2nd IASTED International Conference Visualization, Imaging and Image Processing, Malaga, Spain, 9–12 September, 2002. [4] R. Radha, S. Jeyalakshmi An effective algorithm for edges and veins detection in leaf images IEEE World Congress on Computing and Communication Technologies (2014), pp. 128-131 [5] Park J, Hwang E, and Nam Y, “Utilising venation features for efficient leaf image retrieval,” The Journal of Systems and Software, 2008, 81: 71–82. [6] Jin Yingen. Botany. Beijing: Science Press, 2006. [7] Li Y.F., Zhu Q.S., Cao Y.K., Wang C.L, “A Leaf Vein Extraction Method Based On Snakes Technique,” In: International Conference on Neural Networks and Brain 2005, vol. 2, 13–15 Oct. 2005, pp. 885–888. [8] Clarke, J. ,Barman ,S. ,Remagnino ,P. ,Bailey ,K. ,Kirkup ,D. ,Mayo ,S. ,Wilkin , P.: Venation pattern analysis of leaf images. Lecture Notes In Computer Science 4292, 427–436(2006) [9] H. Fu, Z. Chi A two-stage approach for leaf vein extraction Proceedings of the IEEE International Conference on Neural Networks and Signal Processing, 1 (2003),pp. 208-211 [10] T. Chan, L. Vese An Active Contour Model Without Edges Scale-Space Theo-ries in Computer Vision (1999), pp. 141-151 [11] I. Philipp, T. Rath Improving plant discrimination in image processing by use of different colour space transformations Comput. Electron. Agric., 35 (1) (2002), pp. 1-15 [12] X. Zheng, X. Wang Leaf vein extraction based on gray-scale morphology International Journal of Image Graphics Signal Process., 2 (2) (2010), p. 25 [13] D.S. Bright, E.B. Steel Two-dimensional top hat filter for extracting spots and spheres from digital images J. Microsc., 146 (2) (1987), pp. 191-200 [14] H. Fu, Z. Chi Combined thresholding and neural network approach for vein pattern extraction from leaf images IEE Proc. Vis. Image Signal Process., 153 (6) (2006), pp.881-892 [15] Nam Y., Hwang E., and Kim D., “A similarity-based leaf image retrieval scheme: Joining shape and venation features,” Computer Vision and Image Understanding,2008, 110 (2): 245-259. [16] A. Salima, Y. Herdiyeni, S. Douady Leaf vein segmentation of medicinal plant using Hessian matrix IEEE International Conference on Advanced Computer Scienceand Information Systems [17] K.B. Lee, K.S. Hong An implementation of leaf recognition system using leaf vein and shape Int. J. Bio-Sci. Bio-Technol., 5 (2) (2013), pp. 57-66 [18] Soille P., “Morphological image analysis applied to crop field mapping,” Image and Vision Computing, 2000, 18: 1025 1032 [19] M.G. Larese, R. Nam´ ?as, R.M. Craviotto, M.R. Arango, C. Gallo, P.M. Granitto Automatic classification of legumes using leaf vein image features Pattern Recognit.,47 (1) (2014), pp. 158-168 [20] T. Ojala, M. Pietik¨ ainen, and D. Harwood (1994), “Performance evaluation of texture measures with classification based on Kullback discrimination of distributions”, Proceedings of the 12th IAPR International Conference on Pattern Recognition (ICPR 1994), vol. 1, pp. 582–585. [21] T. Chokey and S. Jain, \"Quality Assessment of Crops using Machine Learning Techniques,\" 2019 Amity International Conference on Artificial Intelligence (AICAI), Dubai, United Arab Emirates, 2019, pp. 259-263.doi: 10.1109/AICAI.2019.8701294 [22] ?N. Shah and S. Jain, \"Detection of Disease in Cotton Leaf using Artificial Neural Network,\" 2019 Amity International Conference on Artificial Intelligence (AICAI), Dubai, United Arab Emirates, 2019, pp. 473-476. doi: 10.1109/AICAI.2019.8701311 [23] S. Singh and S. Jain, \"Detection and Classification of Plant Disease Using Artificial Intelligence,\" 2024 11th International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions) (ICRITO), Noida, India, 2024, pp. 1-5, doi: 10.1109/ICRITO61523.2024.10522449. [24] Jain, Sarika& Ali, Kunwar. (2022). Brinjal Disease Classification Using Deep Learning. 1-5. 10.1109/ICRITO56286.2022.9965112.

Copyright

Copyright © 2025 Nikunj Babbar, Sarika Jain, Ajay Vikram. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.