Enhanced Image–Based Detection and Segmentation of Plant Leaf Diseases Using Grayscale Conversion and Thresholding Techniques

Plant phenotyping has become a pivotal aspect of modern agriculture, providing tools for monitoring plant health and improving crop yield. This study introduces an efficient and cost-effective image processing approach for detecting and segmenting healthy and diseased plant leaves. The methodology focuses on three primary stages: preprocessing, segmentation, and post-processing. During preprocessing, the acquired leaf images are enhanced using grayscale conversion to eliminate noise and simplify image analysis. In the segmentation phase, thresholding techniques are applied to isolate the leaf from its background, enabling more accurate disease identification. The post-processing stage involves refining the segmented leaf images, detecting edges, and analyzing contours to better understand the shape and structure of the leaves. To evaluate the proposed method, a dataset consisting of 100 leaf images was utilized. Experimental results demonstrated that the algorithm successfully detected 68% of infected leaves and 75% of healthy leaves, achieving an overall accuracy of 66.66%. These findings highlight the potential of grayscale conversion and thresholding techniques in automating disease detection, particularly in resource-limited agricultural environments. While the method provides promising results, limitations such as dataset diversity and reliance on grayscale processing were identified. Future enhancements will include the incorporation of color information, the application of advanced machine learning algorithms, and the use of larger, more diverse datasets to improve accuracy and robustness. Overall, this study presents a foundational step toward the development of automated systems for efficient plant disease monitoring and management in precision agriculture.