常用边缘检测技术的对比

边缘检测是图像处理和计算机视觉中的基本问题,边缘检测的目的是标识数字图像中亮度变化明显的点.经典的边缘检测算法如canny算子等是通过计算图像中局部小区域的差分来实现边缘检测的.这类算子对噪声非常敏感,并且常常会在检测边缘的同时加强噪声.多尺度形态学边缘检测利用不同的结构元素去作用图像,通过形态腐性和形态膨胀操作,获得了效果很好的图像边缘检测算法.单尺度形态学梯度算子也能很好提检测图像边缘,但结构元素的选取对输出结果影响较大.通过使用多尺度形态学梯度算子,可以弥补结构元素的大小问题.仿真结果表明,该算法能得到较为理想的图像边缘信息,其抗噪声性能明显优于经典的算子检测算法.

Comparison of common edge detection technologies

Edge detection is the basic problem of image processing and computer vision, the purpose of which is to identify the points with obvious change in brightness in digital images. The classical edge detection algorithms, such as canny operator, etc, realize edge detection through calculating the difference of local small area in the image. This kind of operator is sensitive to noise and often strengthen the noise in edge detection. Multi- scale morphological edge detection technology deals with images by using different structure elements , and an image edge detection algorithm with better effect is obtained through the operations of form corrosion and morphological dilation. Single-scale morphological gradient operator is very good at image edge detection too, but the outcome is influenced by the selection of structure elements greatly.. By using multi-scale morphological gradient operator, the size of structural elements can be remedied. The simulation results show that the algorithm can obtain ideal image edge information and its anti-noise performance is superior to classical operator detection algorithm.