An improved FCM algorithm for ripe fruit image segmentation

Fuzzy C-Means (FCM) clustering algorithm has been widely used in the field of image segmentation with its good clustering efficiency. However, it may cause a time-consuming result and even convergence to local minima because of its local search character while with an improper initial value. Therefore, an improved FCM algorithm is proposed in this paper to solve the ripe fruit image segmentation problem. In the proposed approach, the concept of the neighborhood density is introduced to initialize the cluster center to avoid the improper initial value, which is based on the neighborhood correlation of the data space. Then a resample image method based on entropy constraint is used to reduce the data set, so that the clustering time will be reduced. The effectiveness and efficiency of the proposed approach are demonstrated by experimental studies with some standard data sets and real tomato images.