基于形态学梯度矢量的图像边缘提取算法

Edge Segmentation Algorithm Based on Morphological Gradient Vector

图像的边缘在很大程度上可以用梯度的概念来解释和描述,而现有的形态学梯度边缘检测算子抹煞了梯度的矢量性。文章提出了一种新的图像边缘提取算法:在边缘检测部分提出了具有方向估计的形态学梯度算子,且从理论和实际应用两个方面给予证明。并将模糊处理加入该系列算子,使这些算子在噪声抑制和提高边缘清晰度两方面均有较好的表现。同时在图像分割部分改进了最佳阈值化分割,利用小范围的边缘梯度各方向上的最佳阈值化进行调整,使图像的边缘更加完整、清晰。

The image edge is explained by the gradient. As a vector variable, the gradient has two parts： the magnitude and the direction. The morphological gradient operator, i.e. a popular edge detection operator can detect only the magnitude of the image edge and cannot detect the direction of the image edge, thus lost the information of the edge gradient. This paper presents a new gray level morphological gradient method. The method points out that there is a morphological gradient operator with the direction estimate on the edge detection. The algorithm is validated theoretically and experimentally. The fuzzy process is added into the serial operators, so the noise in the image can be controlled and the clarity of the image edge be increased. Meanwhile, the optimal threshold segmentation is improved by adjusting the optimal threshold values of different directions.