摘要： 该文介绍一种新的分割算法-SUSAN算子。SUSAN算子是一种基于图像局部灰度特征的算法,利用一个圆形的模板对图像进行扫描,比较模板内部的点与模板中心点的灰度值,如果灰度差值小于一定的阈值,就认为该点与中心点的灰度相同。统计模板内部与中心点灰度相同的点的个数,与一个阈值进行比较,判断该点是否属于某个区域的边缘点,从而实现对图像的分割。应用SUSAN算子对苹果图像的缺陷区域进行分割,可以快速准确地分割出苹果图像上轻微的损伤。经过对100幅图像的检验测试,得出分割准确率为96％。

关键词: 苹果, 最小吸收核同值区, 图像, 阈值, 缺陷

Abstract: This paper represents a new thresholding segmentation method——SUSAN operation. It is the method based on the local gray value character of an image ,using a circular mask processing every pixel, comparing the gray value character of every pixel in the mask with the center pixel, if the difference is less than a certain value, the pixel can be considered the same gray value with the center pixel, accounting the number of these pixel, comparing it with a certain value, judging the piexl is the edge of a certain area or not, and carrying out the segmenting process. Using this method, it can segment the slight defect rapidly and exactly. After 100 apple images test, the correct rate is about 96%,

中图分类号: 

• TP391.41