基于数学形态学的自适应边缘检测新算法

Adaptive algorithm of edge detection based on mathematical morphology

考虑到在形态学中,不同形状的结构元素和不同尺度的元素在去噪声和保持图像细节方面的作用是不同的,提出了一种基于多结构多尺度自适应形态边缘检测算法。对一般边缘检测算子做了改进,增加了边缘细节信息。通过计算检测后的边缘信息熵,自适应确定权值系数。将多结构元素检测的边缘和多尺度元素检测的边缘做融合处理,得到最终的图像边缘。实验结果表明,与几种经典边缘检测算法相比,所提出的算法能有效地抑制图像的多种噪声对边缘检测的影响,较好地保持图像边缘细节,自适应提取完整连续边缘。

As different structure elements and different scale elements play different roles in removing noise and retaining image details, an adaptive algorithm of edge detection based on multiple structure and muhi-scale elements was presented. Firstly, general edge detectors were improved to increase the details of image edge. Secondly, the weights were determined adaptively by calculating the entropies of the edges-detected. Finally, the edges of muhiple structure elements detection and the ones of multi-scale elements detection were fused to achieve the final image edges. The experimental results show that the proposed algorithm, compared with a few classic algorithms of edge detection, can effectively eliminate the influence of noise variety on the edge detection, better retain the image details, and adaptively detect the complete and successive edges.