Leaf segmentation is a crucial step in agronomy applications such as leaf counting, plant disease detection, phenotyping, and growth analysis, which helps farmers in plant breeding, plant health monitoring, horticulture, and improving crop yield. The problem of poor segmentation results often pose a significant challenge in the context of early detection of plant diseases. Traditional methods suffer from poor segmentation results and lack of accuracy. This work presents leaf segmentation using image processing techniques, segmenting the leaf from its background with K-means clustering, and optimizing the threshold with Logistic particle swarm optimization (PSO) which enhances the local search efficiency to address these challenges. The methodology entails taking 10 images of different classes from the PlantVillage dataset, applying contrast adjustment, and then reshaping. The preprocessed images are converted from RGB to \({{L}^{*}}\,{{a}^{*}}\,{{b}^{*}}\) color space, since it is device-independent and efficient in distinguishing subtle color differences, which enhances the segmentation accuracy. The leaf picture is first segmented into two clusters using K-means clustering, and these are subsequently optimized using Logistic Particle Swarm Optimization. The proposed method is compared with three algorithms to validate the performance. The results show significant change in terms of segmentation accuracy which demonstrate the effectiveness of the proposed method in addressing the inefficiencies in leaf segmentation.