文档介绍：摘要在图像理解、图像匹配、三维重建及模式识别等领域中,特征点的检测具有十分重要的意义。特征点在保留图像中物体的重要特征信息的同时有效地减少了信息的数据量,使得对图像处理时运算量大大减少。特征点的的定义有多种不同的表述,如图像中灰度值和像素剧烈变化的点、图像边界上具有较高曲率的点等。对于特征点的定义决定了特征点的特性,同时也决定了所检测出的特征点的检测所采用的方法。 SUSAN (Smallest Univalue Segment Assimilating Nucleus) 算法是上世纪九十年代中期提出的一种边角点检测算法,该算法是一种基于图像局部灰度特征, 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,否则就认为该点与中心点有明显的差异。统计模板内部与中心点灰度相同的点的个数, 与一个全局阈值进行比较,判断该点是否属于角点,从而实现对图像特征点的检测。本文采用 matlab 语言实现了 SUSAN 特征点监测并利用模拟图像和真实图像对算法进行了验证,实验表明 SUSAN 特征点检测算法能有效提取图像中的特征点。与传统的角点检测算法相比, SUSAN 算法是一种基于图像灰度比较的算法,不涉及梯度的计算。具有速度快、抗干扰能力强等特点。关键字: SUSAN 算法边缘检测角点检测 matlab Abstract In image understanding , image matching , reconstruction , and pattern recognition , the feature point detection isof great points of objects while retaining the important features of the image information , while effectively reducing the amount of data information , making the image processing operations is greatly points , there are different definitions of expression , if the image has a gray value and dramatic changes in the point image pixel on the boundary of high curvature the definition of the feature points of the feature points of the characteristics of the decision ,but also determines the feature points detected in the detection method used. SUSAN(Smallest Univalue Segment Assimilating Nucleus) algorithm is the mid ies of last century a corner detection algorithm , which is based on local gray level image features , using a circular template to scan the image , comparetemplate within the template center point and the gray value , if the gray level difference is less than a certain threshold , to that point and the center of the gray level of the same , otherwise the point that there are significant differences with the center gray inside and the same template the number of dots , with a global threshold value parison to determine whether the corner point , enabling the detection of image features.