A Comprehensive Framework for Leaf Shape Analysis: From Preprocessing to Morphology-Based Classification

This article described a deep learning and morphology-based framework for automatically identifying plant species using smartphone leaf images. The framework includes a robust background removal process, preprocessing images (using CLAHE, applying a blur to create an area of effect, implementing adaptive thresholding), morphological refinement, and employing a hybrid classifier (CNN + MobileNetV2). A web-based dashboard built with Flask provides real-time predictions, morphological measurements of the plant species, radar charts, and visualisation of the entire image processing pipeline. Experimental results obtained from 19 different plant species demonstrated that the developed system yielded 74% accuracy in identifying the genus and 69% accuracy in identifying the species, while processing each leaf image required an average of 1.5 seconds. As a result of the morphological refinement processes, the average Hausdorff distance for boundary noise was decreased by 62%; thus, the model achieved accurate classifications with confidence levels of 71% to 76% for Betel, Tulsi, Mango, and Neem plants. There were still several cases in which the developed model could not accurately identify a species because of several shared characteristics: shared vein patterns (Betel vs. Arali vs. Amruthaballi), damaged leaves, or overlapping leaves. Overall, the proposed framework is a practical and effective approach for automated plant identification by combining traditional image processing techniques (i.e., morphological analysis with deep learning techniques (transfer learning). Future research to enhance the robustness of the classification process will focus on increasing the diversity of the datasets and developing better segmentation algorithms.

1. Ming-Kuei Hu, “Visual pattern recognition by moment invariants,” IRE Transactions on Information Theory, vol. 8, no. 2, pp. 179–187, 1962. DOI: 10.1109/TIT.1962.1057692 2. Faramarz Mokhtarian and Alan K. Mackworth, “A theory of multiscale, curvature-based shape representation for planar curves,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 14, no. 8, pp. 789–805, 1992. DOI: 10.1109/34.149591 3. J. Du et al., “Leaf shape based plant species recognition,” Applied Mathematics and Computation, vol. 185, no. 2, pp. 883–893, 2007. DOI: 10.1016/j.amc.2006.07.072 4. S. G. Wu, F. S. Bao, E. Y. Xu, Y.-X. Wang, Y.-F. Chang and Q.-L. Xiang, “A leaf recognition algorithm for plant classification using probabilistic neural network,” Proc. IEEE International Symposium on Signal Processing and Information Technology (ISSPIT), 2007, pp. 11–16. DOI: 10.1109/ISSPIT.2007.4458016 5. J. S. Cope et al., “Plant species identification using digital morphometrics: A review,” Expert Systems with Applications, vol. 39, no. 8, pp. 7562–7573, 2012. DOI: 10.1016/j.eswa.2012.01.073 6. Abdul Kadir et al., “Leaf classification using shape, color, and texture features,” International Journal of Computer Trends and Technology, vol. 1, no. 3, pp. 225–230, 2011. DOI: 10.5121/sipij.2011.2301 7. Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton, “ImageNet classification with deep convolutional neural networks,” Advances in Neural Information Processing Systems (NeurIPS), 2012. DOI: 10.1145/3065386 8. Olaf Ronneberger, Philipp Fischer, and Thomas Brox, “U-Net: Convolutional networks for biomedical image segmentation,” Medical Image Computing and Computer-Assisted Intervention (MICCAI), pp. 234–241, 2015. DOI: 10.1007/978-3-319-24574-4_28 9. Gao Huang et al., “CondenseNet: An efficient DenseNet using learned group convolutions,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018. DOI: 10.1109/CVPR.2018.00320 10. E. Adetiba et al., “LeafsnapNet: A deep convolutional neural network for plant classification,” International Journal of Advanced Computer Science and Applications, vol. 12, no. 6, pp. 1–8, 2021. DOI: 10.3844/jcssp.2021.349.363