susan的理论知识



如图1所示，图片是白色背景，有一个颜色比较暗淡的矩形。在图片上有5个圆形区域。圆形区域表示的是掩码区域。把圆形区域内的每一个位置的像素值与圆心处的像素值相比较，那么圆中的的像素可以分为两类，一类是像素值与圆心处的像素值相近的，另一类是像素值与圆心的处的像素值相差比较大的。那么，把第一类（像素值与圆心处的像素值相近的）所组成的区域称为USAN，USAN的全称是“Univalue Segment Assimilating Nucleus”。图2 中显示了图1中的5个圆的USAN区域。图2中，圆中的白色区域是表示该掩码区域的USAN区域。

USAN区域表征了图像局部结构的信息，对USAN的区域的大小及二阶矩的分析，可以得到图像中的角点，边缘等信息-susan theoretical knowledge in Figure 1, the background image is white, there is a color in dim rectangle. On the image there five circular area. A circular area represents the area of ​ ​ the mask. The circular area of ​ ​ each position and the center pixel value compared to the value at the pixel, then the circle of the pixels can be divided into two categories, one is at the center pixel value and the pixel values ​ ​ are similar, and the other is the pixel value of the pixel values ​ ​ at the center of the relatively large difference. So, the first category (the center pixel value and the pixel values ​ ​ at similar) consisting of area is called USAN, USAN' s full name is " Univalue Segment Assimilating Nucleus" . Figure 2 shows the Figure 1 USAN circle area 5. Figure 2, the circle represents the white area is the area of ​ ​ the mask region USAN. USAN area characterizes the local structure of the image informat