Abstract:
Aiming at the defect that Fuzzy C-Means(FCM) algorithm is sensitive to noise, this paper proposes a FCM image segmentation method based on Total Variation(TV) model by introducing TV function into FCM objective function. The method can adaptively modify penalty factor of image fidelity term according to the texture variation of image. In consideration of the segmentation cost, the noise of image is smoothed within the iteration. Experimental results show that the proposed method can segment the image effectively and solve the confliction between the noise smoothing and the accurate segmentation.
Key words:
image segmentation,
Fuzzy C-Means(FCM),
Total Variation(TV)
摘要: 针对模糊C均值(FCM)算法对噪声敏感的缺点,在FCM目标函数中引入全变分惩罚函数,提出一种基于全变分模型的FCM图像分割方法。该方法根据图像的纹理变化,自适应调整图像保真项的惩罚因子,同时在考虑分割代价的情况下,使迭代循环过程中的图像噪声得到平滑。实验结果表明,该方法能提高图像的分割效果,有效解决噪声抑制与精确分割之间的矛盾。
关键词:
图像分割,
模糊C-均值,
全变分
CLC Number:
BANG Dai-Jiang, DU Feng-Fei, LIN Yao-Quan. Adaptive FCM Image Segmentation Based on Total Variation Model[J]. Computer Engineering, 2010, 36(11): 203-205,208.
彭代强, 杜鹏飞, 林幼权. 基于全变分模型的自适应FCM图像分割[J]. 计算机工程, 2010, 36(11): 203-205,208.