基于多尺度轮廓结构元素的多形状边缘检测

在图像边缘检测过程中,针对滤除噪声及有效保留图像边缘信息这对矛盾点进行了研究,给出一种基于多尺度轮廓结构元素的多形状边缘检测算法。该算法通过多次使用轮廓结构元素的开最大和闭最小运算操作滤除噪声,运算次数通过比较图像峰值信噪比确定,降低结构元素对边缘信息的影响;然后采用多形状多尺度结构元素提取图像边缘,并利用图像峰值信噪比控制结构元素尺度的选取。与经典边缘检测算法相比,该算法具有更强的去噪声能力,且能保留更多的图像细节。仿真实验表明,有区别地使用轮廓结构元素及多形状多尺度结构元素,能有效去噪并保留边缘信息。

Morphologic edge detection based on multi-scale contour structuring elements with multiple structuring elements

Based on the research about how to keep more edge information effectively while filtering the noise in the process of image edge detection, this paper proposed an edge detection algorithm based on multi-scale contour structuring elements with multiple structuring elements. Firstly, it adopted a class of open maximum and close minimum morphological filter by using multi-scale contour structuring elements several times determined by the peak signal noise ratioPSNR, thereby reducing it on the impact of the edge. Then, it adopted multiple structuring elements to extract the edge, determined the scale of the multiple structuring elements by the PSNR. Compared with the classical edge detection algorithm, this algorithm had robust ability to filter the noise, and could retain more edge information. Experimental results show that using the multi-scale contour structuring elements and the multiple structuring elements differently can filter the noise effectively and retain the edge information.