基于多重分形滤波的遥感图像岩性信息增强研究

Research on the Strengthen of Lithologic Information of Remote Sensing Image Based on Multi-Fractal Filter

近年来,随着遥感技术和地质学的深入发展,越来越多学者尝试利用遥感数据进行岩性信息增强的深入研究。但在茂密的植被覆盖地区,岩石出露有限,地层可见度较低,不易在遥感图像上直接提取分析,使得传统的处理方法遇到了挑战。本文首先从空间域反射波段的光学遥感图像入手,选择蚌埠涂山二长花岗岩、大别山二长花岗岩、淮南上窑寒武系馒毛组页岩和凤阳韭山馒毛组页岩四幅遥感影像子图像作为研究对象,利用傅里叶变换将遥感图像变换到频率空间,得到能量分布谱确定分形滤波器的阈值,借助这些阈值在频率域上运用不同的滤波器,并将其变换到空间域上以实现将异常和背景进行分离,达到遥感图像岩性信息增强的目的。

In recent years, with the further development of remote sensing technology and geology, more and more scholars have tried to use remote sensing data to enhance the lithologic information in-depth study. However,in thick vegetation areas, Limited rocky outcrops and weak stratigraphic visibility make it difficult to directly extract and analyze information in remote sensing image, and therefore it is a challenge for traditional processing methods. Starting from the optical remote sensing images of the spatial domain reflection band, and choosing four sub images, such as the monzonitic granite of Bengbu Tushan and Dabie Mountain and the shales in the Manmao Formation of the kiln of Cambrian system in Huainan Province and that of Fengyang Jiushan, as the research object, this paper transforms the remote sensing image to the frequency space by using Fourier transformation, to obtain the energy distribution of spectrum to determine the threshold of the fractal filter. We apply different filters on the frequency domain with the help of these thresholds. Finally we transform it into spatial domain in order to separate anomaly and background and to improve the lithologic information of remote sensing image.