边缘检测的形态学算法与传统算法比较

The Comparison Between Morphology Algorithm and Traditional Algorithms for Edge Detection

由于图像的边缘通常含有大量重要信息,因此,边缘检测成为图像处理的一个重要环节,其检测算法也获得了广泛的研究,已经形成了Roberts、Laplacian、Canny等多种算法。但这些传统算法在边缘检测精度和抗噪声性能方面还存在一定的问题。文章运用数学形态学边缘检测算法的结构元素变换,对无噪声图像检测出多幅边缘图;对噪声图像采用改进的开启运算,先用3×3的结构元素进行腐蚀,后用5×5的结构元素进行膨胀,用边缘检测算子f°B-f进行检测,并与传统算法和不变结构元素的形态学开启运算的结果进行了比较。实验结果表明,灵活多变的数学形态学边缘检测算法在检测精度和抗噪声性能上都优于传统算法。

The edge of a picture generally implies much useful information so the edge detection is important in image processing and the detection algorithm is studied extensively. There have been various algorithms such as Roberts, Laplacian, Canny and so on. But the traditional algorithms have drawbacks in the detection accuracy and the anti -noise performance. In this paper, the mathematical morphology detection algorithm with the construction element transformation is used to detect a noiseless picture. The improved open calculation, that is using 3 x 3 construction element to decay, then using 5 x 5 constructionelement to inflate and finally using operator foB -f to detect, is used to the noise picture. The morphology algorithm with construction element transformation is compared with the traditional algorithms and the constant construction element morphology algorithm. The results show that the flexible mathematicalmorphology edge detection algorithm is better than the traditional algorithms in both detection accuracy and anti - noise performance.