基于向量排序的图像边缘提取方法

Image edge detection using new method based on vector order statistics algorithm

在传统的图像边缘提取的过程中,往往只是注重像素点的灰度值,忽视了相邻像素点间的灰度跃变的方向信息,从而导致了得到的图像边缘的连接性不好,而且缺失了许多重要的细节。结合向量排序统计提出了一种新的边缘提取的方法,利用该方法分别得到图像R,G,B三分量的边缘强度,同时应用非极大抑制和自适应阈值对所得图像边缘进行处理,最后将处理后的三分量边缘信息融合,得到了最终的边缘图像。实验结果表明,该方法很好的保证了图像边缘的连续性和准确性。

During the traditional process of color image edge detection, some methods are mostly concerned on the pixels＇ grey level but ignored the grey level＇s changing direction. So they result in the thick image edge and the deficiency of continuity, thus, the final edge map can not describe some important details. Based on the vector order statistics operators, a new color image edge detection method is developed, using this method, firstly the R, G, B components＇ edge intensity is got respectively, then these information is analyzed by implying the non-maxima suppression and adaptive threshold setting, after all the pixels are processed, the method integrate the three components＇ information and the edge map are generated. Simulation results show that this algorithm guarantees the color image edge＇s continuity and accuracy successfully.