Implementation of Watershed Algorithm for Rice Plant Image Segmentation to Analyze Growth Levels of Plant Image
Teri Ade Putra, Yuhandri, Agung Ramadhanu · 2025
This study aims to implement the Watershed algorithm with an enhanced approach by incorporating a Gradient parameter based on distance and color intensity to improve the segmentation quality of rice plant images. The dataset used in this research consists of 300 images of rice plants, with 200 images used for training and 100 images for testing. The research methodology involves several stages of image preprocessing, including image segmentation, grayscale conversion, filtering, Otsu thresholding, normalization, color segmentation, and edge detection. Following these preprocessing steps, the Watershed algorithm is applied. To evaluate the performance of the proposed approach, standard classification metrics such as accuracy, precision, and recall are employed. The experimental results demonstrate that the Watershed algorithm, when implemented with the Gradient parameter, significantly enhances segmentation accuracy compared to the conventional Watershed method. The enhanced algorithm achieved an average segmentation accuracy of 90%, whereas the conventional algorithm reached only 80% on average. This improvement indicates that the Gradient parameter effectively enhances image segmentation in scenarios involving varying pixel intensities and the presence of noise. However, the proposed method may encounter challenges in real-time applications due to its computational complexity, and its performance may vary under extreme lighting conditions or with different plant species. In conclusion, the study confirms that the integration of the Gradient parameter into the Watershed algorithm offers an effective solution for improving segmentation accuracy in rice plant image analysis. These findings support the potential deployment of automated monitoring systems in smart farming applications, while also opening new opportunities for adaptation to other plant types and various environmental conditions.