Pomegranate MR image analysis using fuzzy clustering algorithms

Abstract

In this paper, the process of the pomegranate magnetic resonance (MR) images was studied.  Its internal structure is composed of tissue and seeds, which indicate the dependency between the maturity and internal quality.  The latter properties are important in pomegranate’s sorting and cannot be measured manually.  In this paper, an automatic algorithm was proposed to segment the internal structure of pomegranates.  Since the intensities of the calyx and stem of the pomegranate MR image are closely related to that of the soft tissue, their corresponding pixels are therefore labeled in the same class of the internal soft tissues.  In order to solve this problem, the exact shape of the pomegranate is first extracted from the background of the image using active contour models (ACMs).  Then, the stem and calyx are removed using morphological filters.  We have also proposed an improved version of the fuzzy c-means algorithm (FCM), the spatial FCM (SFCM), for segmentation of MR images of pomegranate.  SFCM is realized by incorporating the spatial neighborhood information into the standard FCM and modifying the membership weighting of each cluster.  SFCM employs spatial information of adjacent pixels leading to an improvement of the results.  It thus outperforms other techniques like FCM, even in the presence of Gaussian, salt and pepper, and speckle noises.

Keywords: MRI, pomegranate, image segmentation, spatial fuzzy c-means, morphological filter