传统的聚类对像素点归类的方法仅利用了像素的灰度信息，而忽视空间位置信息

The traditional clustering method of categorizing pixels only uses the gray information of the pixels, while ignoring the spatial location information. At the same time, the Euclidean distance measure is used to measure the similarity between pixels. However, in real image segmentation, the gray values ​​of the pixels often have little difference between the target objects, and it is even difficult to distinguish the boundaries of the objects. At this time, the problem of pixel classification in the boundary area is an uncertainty problem. For such problems It is often difficult to determine whether it is the distribution function of a random process or the membership function of fuzzy mathematics.

The traditional clustering method of classifying pixels only utilizes the grayscaness information of pixels, neglects spatial location information, and uses Eucalypter distance measurement to measure the similarity between pixels. However, in the reality image segmentation, the difference between the target objects is often not large, or even difficult to distinguish the boundaries of the object, at this time, the boundary area of the pixel classification problem is a problem of uncertainty, for such problems, whether it is the random process distribution function, or fuzzy mathematical membership function is often not easy to determine.

传统的聚类对像素点归类的方法仅利用了像素的灰度信息，而忽视空间位置信息，同时采用欧氏距离测度来衡量像素之间的相似度。然而，在现实图像分割中，目标对象之间往往像素灰度值相差不大，甚至很难区分对象的边界，此时，边界区域的像素点分类问题是一个不确定性问题，对于此类问题，无论是随机过程的分布函数，还是模糊数学的隶属函数往往都不易确定。<br>