基于二维灰度直方图的最小模糊熵分割方法

Minimum fuzzy entropy image segmentation based on 2D-gray level histogram

本文在一维最大模糊熵分割方法的基础上,根据图像目标和背景内部像素灰度值的一致性和集中性,提出了一种新的图像分割隶属度函数,从而得到最小模糊熵分割方法.本文还针对传统的基于一维灰度直方图的模糊熵分割方法不能反应图像的空间信息,抗噪声能力差的缺点,提出了基于二维灰度直方图的模糊熵分割算法.本实验结果证明,最小模糊熵分割方法对于某些图像的分割效果要好于最大模糊熵分割效果,而二维分割方法对于绝大多数图像,都具有很强的鲁棒性和抗噪能力,分割效果明显优于一维的方法,而且方便地推广到其他的一维熵分割方法中.

Based on the 1D-Maximum Fuzzy Entropy image segmentation, a new function for image segmentation is presented according to the consistency and concentricity of the pixel's gray levels inside the image objects and the background, the minimum fuzzy entropy image segmentation is then achieved. aiming at the drawbacks of being unable to reflect the space information and the low ability of resisting yawp, a fuzzy entropy segmentation algorithm based on the 2D-gray level is put forward. The experimental result shows that the minimum fuzzy entropy image segmentation is more fitting for some images than maximum fuzzy entropy image segmentation, and the 2D-segmentation has the very high ability of resisting noise and the LuBang attribute, the segmentation effect obviously excel than 1D-segmentation, further more it is convenient to popularize to other 1D-Entropy segmentation.