基于四叉树和曲线拟合的FCM图像分割方法

针对模糊C均值聚类算法对大数据量的情况进行分割时存在耗时过长的问题，提出基于改进四叉树的FCM分割方法。该算法首先求取图像的四叉树，将四叉树节点集合作为模糊聚类算法的样本集，减小样本空间，然后利用FCM算法实现图像的边缘分割；针对分割处理后的图像带来的锯齿状边缘，再利用曲线拟合算法对图像边缘进行平滑处理。在标准摄影师图像和肝脏核磁图像的数据集上实验，实验结果可以看出算法提高了FCM的运算速度，对医学图像具有更高的分割精度，便于为后续诊断和数据分析提供基础。

Method of FCM Image Segmentation Based on Quadtree and Curve Fitting

The improved quadtree is applied in image segmentation for fuzzy C-means clustering algorithm to solve the problem of long time consuming .The algorithm firstly obtains the image of quadtree, using the quadtree node collection as a sample set of fuzzy clustering algorithm and reducing the sample space, then uses FCM algorithm to realize the image segmentation; With the jagged edges for the image segmentation brings, it uses curve fitting algorithm for image edge smoothing.The algorithm is applied in standard photographer image and liver magnetic resonance image. Experiment results show that this algorithm improves the segmentation speed of FCM and it has higher accuracy of segmentation of medical image and provides a basis for further diagnosis and data analysis.