摘要
提出了一种基于统计的图像边缘检测方法。它首先通过计算图像中每一点的梯度向量以及方差—协方差矩阵,然后对每一点的梯度向量进行非极大值抑制,最后使用统计的方法求出局部标准化梯度阈值来检测图像的真实边缘。通过与经典的边缘算子相比较,该方法不仅能很好地检测出图像的真实边缘,而且有效地抑制了虚假边缘的产生,取得了良好的效果。
A statistical approach of edge detection was presented. First, this approach determined gradient vector and variancecovariance. Second,it made use of non-maxima suppression to every gradient vector. Last, local standardization for gradient threshold base on statistical approach was computed and use it for edge detection. To compare with well-known edge detectors, this approach not only detects real boundary of image, but also effectively suppresses the false edge of image. This new approach receives a good effect.
出处
《计算机应用研究》
CSCD
北大核心
2007年第8期204-205,208,共3页
Application Research of Computers
基金
国家自然科学基金资助项目(40401003)
关键词
局部标准化阈值
图像平滑
边缘检测
非极大值抑制
local standardization for gradient threshold
image smoothing
edge detection
non-maxima suppression
