基于粒度分层技术与FCM算法的图像分割

Study on image segmentation based on granularity stratification technique and FCM algorithm

文章提出了一种高效快速地实现图像分割的方法。该方法采用自适应像素梯度法进行图像预处理,能有效地消除噪声,保留细小纹理,突出图像边缘。该方法基于粒度分层技术与模糊C均值聚类算法实现图像分割,基于粒度分层技术进行粗粒度划分,得到最佳粗粒度层;在此层上进行FCM算法,通过建立目标函数,构建模糊矩阵,确定聚类中心,实现一系列迭代优化;最后进行细粒度划分,并选出最佳细粒度层,达到目标图像与背景分离的分割效果。仿真实验证明分割效果高效且快速。

This paper presents a fast and efficient way to achieve image segmentation. This method adopts adaptive pixel gradient method for image preprocessing, which can effectively eliminate the noise, retain fine texture and protrude the edge. The method is based on granularity stratification technique and fuzzy C means （FCM） clustering algorithm to realize image segmentation and get the best coarse grain layer, then based on this layer to carry out FCM algorithm, through set target functions to ensure cluster centers and realize a series of iterative optimization; finally to carry out fine-grained division and select the best fine grain layer, so as to achieve the segmentation results of target image and background segmentation. Simulation results show that the segmentation effects are efficiently and fast.