【摘要】 为提取出更丰富流畅、定位更准确的边缘信息,基于多尺度多方向结构元素,引入信息熵加权系数,改进了形态学边缘检测算法.首先,为去除图像噪声,选定2种不同尺度的结构元素进行形态学开闭运算;然后,利用4种不同方向的结构元素进行形态学边缘检测,可以获得4幅不同方向结构元素下的边缘图像;再根据每幅边缘图像的信息熵来确定权值,并将这些边缘图像按照比例进行加权求和即可得到较为理想的边缘图像.仿真实验证明:改进的形态学边缘检测算法具有很强的抗噪性能,而且检测到的边缘信息更完整流畅,具有一定的适用性和优越性.更多还原

【Abstract】 To extract more abundant and smooth edge information and more accurate positioning, based on multi-scale and multi-directional structural elements, the weighted coefficient of information entropy is introduced to improve the morphological edge detection algorithm. Selecting two different scale structure elements can reduce the noise of the image. Using four different direction structure elements,we can get the edge image of different direction structure elements. According to the information entropy of each edge image, we can determine the weight, and then we can get the ideal edge image by weighted sum of these edge images according to the proportion. Simulation results show that the improved morphological edge detection algorithm has strong anti noise performance, and the edge information detected is more smooth and complete, which has certain applicability and superiority.更多还原