摘要
针对传统Canny算子的高斯滤波参数和高低阈值选择困难,以及会造成缓变边缘丢失和假边缘的现象,提出用广义交叉验证准则进行小波阈值的自适应选取,用此阈值的广义阈值函数的小波滤波方法代替高斯滤波器对含噪图像去噪,然后采用最大类间方差的方法来实现Canny算子高低阈值的自适应选择,并用此高低阈值检测及连接图像的边缘。实验结果表明,改进的算法改善了噪声干扰情况下Canny算子的边缘提取效果,有效提高了边缘检测的准确性。
In order to solve the problems of difficult selecting of Gaussian filter parameter, high and low thresh- old in traditional Canny operator, the missing of slowly varying edge and the producing of feigned edge, an improved edge detection algorithm is proposed. This method selects wavelet threshold using Generalized Cross Validation crite- ria automatically, replaces Gauss filter with generalized threshold function of this threshold to smooth image, adjusts the high and low threshold value automatically by using the Otsu algorithm, and detects and connects edges of the im- age with the threshold. The experiment shows that the reformed algorithm improves the effect of edge detection of Canny algorithm in the case of noise disturbance, and increases the veracity of edge detection effectively.
出处
《计算机仿真》
CSCD
北大核心
2010年第4期252-255,共4页
Computer Simulation
