【摘要】 传统形态学使用固定结构元素对图像进行形态学运算时,容易使图像边缘属性发生改变的。首先计算图像的线性结构张量;然后得到线性结构张量的特征值和特征向量,以此特征确定椭圆结构元素的长短半轴以及方向;最后在构造椭圆结构元素的基础上,定义相应的自适应形态学膨胀和腐蚀以及开闭运算。实验结果表明:利用椭圆结构元素的自适应形态学运算相比传统形态学运算和其他自适应形态学运算,具有较高的边缘保持特性和抗噪性能,梯度幅值相似性偏差和失真较小。更多还原

【Abstract】 Traditional morphology uses fixed structuring elements to perform morphological operations on images,which can easily change the edge of images. Firstly,the linear structure tensor of images is calculated. Then,the eigenvalues and eigenvectors of linear structure tensor are used to determine the semi-majoraxes,the semi-minor axes and the direction of elliptical structuring elements. Finally,on the basis of constructing elliptical structural elements,the adaptive morphological dilation and erosion,opening and closing are defined. The experimental results show that the adaptive morphological operations using elliptical structuring elements have better edge preservation and anti-noise performance than traditional morphological operations and other adaptive morphological operations,and the gradient magnitude similarity deviation is smaller.更多还原