Determining the damage level of leaf borer (Wilsonomyces carpophilus lév.) in some stone fruits using image processing techniques

Agriculture is a critical sector for human survival and plays a major role in the economy, nutrition and environmental protection. However, plant diseases and pests are a major problem that negatively affect agricultural production. Early detection and effective interventions are necessary to reduce the effects of these diseases and increase productivity. In this study, it was aimed to determine the damage level of leaf borer (freckle) disease (Wilsonomyces carpophilus Lév.), which causes crop losses in fruits important for our country, by using image processing techniques. For this purpose, an image processing algorithm was developed by combining K-Means and Gaussian Mixture Models (GMM) clustering algorithms. A dataset containing 200 images of apricot fruit and leaves, cherry and peach leaves, and apricot fruit and leaves with symptoms of leaf borer disease was created. The acquired images were processed with the Image Processing Toolbox module of the MATLAB program and the developed algorithm. Then, the results obtained from the image processing algorithm were compared with expert observation. As a result of the comparison, the overall result for a total of 200 images was that the image processing method predicted the disease levels with an RMSE of 2.86 and Theil UII of 0.1164. The Kolmogorov-Smirnov test was used for the normality assumption of the data, and it was found that the data were normally distributed (p>0.05). The regression coefficient (R² = 0.986, p<0.01) and Pearson correlation coefficient (r = 0.993, p<0.01) between image processing results and expert opinion were found. As a result of the study, it was shown that the developed image processing algorithm can be successfully used instead of expert observation in the diagnosis of leaf borer disease in apricot, cherry and peach fruits.