Analyzing diplocarpon mespili lesions on pear using image processing

Abstract

Pear is an important fruit not only for domestic market but also exportation. During preservation, it is possible to observe developping fungal growth on products arisen from latent infections after taking the products to storages. Under preservation conditions which are suitable for fungal growth, deteriorations and deformations can occur on the products which may cause commercial loses. To prevent these, it is crucial to detect and isolate infected fruits in storages. In this study, a lesion detection algorithm was developed by using image processing methods and support vector machine to prevent infected fruits by Diplocarpon mespili to transmit the disease to healthy ones. With the proposed method, it was found that active disease regions which continue to grow and to increase diameter could be detected with higher accuracy in comparison with passive regions.