WSEAS Transactions on Systems
Print ISSN: 1109-2777, E-ISSN: 2224-2678
Volume 12, 2013
Application of Adaptive MRF Based on Region in Segmentation of Microscopic Image
Authors: , , ,
Abstract: According to the characteristics of microscopic image, an adaptive MRF method based on region is provided for segmentation of microscopic image. Based on a series of filtering and de-noising, morphological gradient is implemented for image. Then a watershed algorithm is used for image’s over-segmentation. In order to reduce the influence of noise, the mean gray value of regional block substitutes for each pixel value in this regional block. The fuzzy c-means algorithm is implemented for initial segmentation of image. In this processing, the mean value and variance of this region is feature value. Then MRF potential function is computed. Condition potential function is represented by membership of regional feature value on clustering center. The connection parameter of priori potential function is adaptively determined according to the connection degree between regional block and its adjacent blocks. The edge of the segmented regions with this algorithm is better than that algorithm in which fixed connection parameter is adopted. Because fuzzy c-means segmentation can get good initial state value, ICM with better real-time is employed to calculate the minimum potential function. Experiments show that this algorithm is better than OTSU, the traditional MRF and regional MRF with fixed connection parameter. This algorithm has better anti-noise ability and edge segmentation image. It has good robustness
Search Articles
Keywords: Segmentation of microscopic image, Markov random field based on region, Watershed, Adaptive connection parameter, image processing, potential function