在形态学梯度边缘检测算子的基础上,综合多结构元和多尺度算法的特性,提出了一种基于小波增强的多结构元、多尺度边缘检测方法,用不同取向的结构元素对图像进行多尺度检测,并综合各尺度下的边缘,得到了噪声存在下的理想边缘。实验表明,本文方法边缘定位准确、轮廓清晰,保留了更多的图像细节,具有较强的抗噪能力。
The multi-structure elements and multi-scale edge detection method is proposed based on image enhanced of wavelet transform. Four elements are used to detect multi-directional edges of image at different scales, The more ideal image edges under the environment of existing noise are obtained by integrating the edge characteristics for various scale. During computation the computation is reduced by simplifying the different scales structuring elements. The experiments show that compared with traditional morphological edge detection algorithms, the method possesses better edge locating ability and keeps the more edge details, but also the ability of noise interference restrained is stronger.