基于多结构元素的遥感图像去噪及边缘检测方法

边缘检测是遥感图像处理的一个重要方面。利用数学形态学原理，提出了一种基于多结构元素的遥感图像边缘检测方法。首先采用基于视觉模型的边缘阈值得到灰度突变像素点，然后利用几种具有代表性的结构元素对这些像素点进行形态学腐蚀操作以区分真实像点和噪声点，将噪声点去除并通过伪中值滤波填补像素值，最后得到图像边缘。实验结果表明，该算法能够在保持图像边缘细节的前提下，很好地滤除图像中的孤立噪声点，在检测精度和抗噪声性能方面都优于传统的形态学算法。

Noise Elimination and Edge Detection Method Based on Multi-Element for Remote Sensing Image

Edge detection is an important aspect for remote sensing image processing. With mathematical morphology principles, a new edge detection algorithm for remote sensing images based on multi-element is presented in this paper. Firstly, The pixels with sharp grey scale are determined by a thresh value based on vision model, and then these pixels are eroded by the special construction elements to distinguish noise dots from real image dots. These noise dots are eliminated and then the pixels are recovered with pseudo-median filtering. Finally the edge picture of the image is acquired. The experimental results show that the new algorithm can powerfully eliminate isolated noise dots of the image at the same time keep the image sharp, and is better than traditional morphology algorithms in aspects of detection accuracy and anti-noise porformance.