FCM算法用于灰度图象分割的研究

FCM Algorithm for the Research of Intensity Image Segmentation

模糊Ｃ均值（ＦＣＭ）算法用于灰度图象分割是一种非监督模糊聚类后再标定的过程，适合灰度图象中存在着模糊和不确定性的特点．但是这种算法存在着一些不足，如聚类数目无法自动确定、运算的开销太大等，因而限制了这种方法的应用．针对这些问题，本文利用直方图分析的方法，自动确定算法的聚类数目和各类的类峰值．并针对ＦＣＭ算法和灰度图象的特点，提出了一种适用于灰度图象分割的快速ＦＣＭ算法（ＱＦＣＭ），使得运算的开销降低，聚类分割的速度显著提高，并从数学和实验上证明了该方法的有效性．

It is a procedure of the label following an unsupervised fuzzy clustering that fuzzyc-means (FCM) algorithm is applied for intensity image segmentation,and it suits for the uncertainand ambiguous characteristic in intensity image. However,there are some deficiencies in the algorithm,for example,the number of clustering can not be determined automatically and the operational cost,for large data sets,is too high,which limit its application. In order to overcome the deficiencies,the number of clustering and the maximum of each cluster are automatically determinedthrough the histogram analysis ; Then, according to the character of FCM and intensity image, aquick fuzzy C-means (QFCM ) algorithm is presented for intensity images segmentation,with whichthe operational cost is reduced and the speed of clustering segmentation is greatly increased. Thefeasibility of the approach is proved by the mathematics and experiment result.