Infected Part Detection based on Affinity Propagation Clustering

Manual identification of defected objects consumes time and effort. These leading researchers try to find out an automatic infected objects detection systems to reduce these denigrate issues, which affects trade business fields, as an example, infected fruits or vegetables in agricultural field. This paper presents an image segmentation method based on affinity propagation (AP) clustering algorithm for detecting infected part in fruits or vegetables. Results show that this methodology is good comparing to K-means algorithm, which gives good results. Nevertheless, AP outperforms that does not need pre-specify cluster number, which is needed in K-means. Some deficiencies occur when using traditional AP, but using sparse version of AP overcomes most of these deficiencies. Extra feature of AP that it works better than K-means as cluster number is increasing or complexity of infected objects is amplifying. AP works