基于模糊算法的遥感图像增强

发布时间：2018-10-31 11:08

【摘要】：“遥感”这个词是由E.L.Pruitt提出的，从20世纪60年代至今已经发展成为一个非常先进的检测技术，在国民经济和国防领域得到了广泛应用，产生了巨大的社会和经济效益，成为了人类从多维和多角度认识宇宙的新方法和新手段。在遥感图像的采集过程中，总是不可避免的受到各种环境因素的影响，从而产生视觉效果差、分辨率低和亮度偏暗等不足，遥感传感器的灰度范围是采集到的遥感图像的灰度所不能完全覆盖的。因此，在提高图像的对比度、突出一些局部细节等方面遥感图像增强技术发挥着积极的作用。图像增强的方法大致可以分为两种：空域图像增强法和变换域图像增强法。空域图像增强顾名思义就是在图像所在的空间域内进行增强处理，换句话说就是直接对图像中每个像素进行某种处理。本文主要介绍一种类似于空域处理的方法--模糊算法，以及一些在传统模糊算法基础上的改进算法。变换域图像增强法是将原始图像变换到特定的变换域内，修改处理变换域内的相关系数来对图像进行相应的增强。图像从空域变换到变换域的方法有傅里叶变换、小波变换、Contourlet变换、NSCT变换、剪切波shearlet变换等。本文将改进的模糊算法分别应用在空域图像增强和变换域图像增强中，提出了两种遥感图像的增强算法：第一：采用改进的模糊算法和一种空域图像增强算法对遥感图像进行增强。由于传统的Pal-King算法阈值难以确定，本文利用改进的OTSU算法来自适应获取阈值，同时由于模糊增强是对全局进行模糊处理的，细节信息增强效果不明显，本章采用相对熵为判别标准，进行领域信息的自适应对比度增强。该算法在保持图像原有亮度的基础上同时增强了图像细节。实验表明，，该算法能够获得很好的视觉效果和更加明显的细节信息。第二：首次将Shearlet变换引入到遥感图像增强中提出了一种基于Shearlet变换的遥感图像增强算法。对Shearlet变换产生的低频系数和各尺度各方向的高频系数分别进行模糊对比增强和改进的模糊增强，再经过Shearlet反变换得到最终的增强结果。实验结果表明，该算法不仅在主观上能获得很好的视觉效果，同时在客观评价中也拥有很好的指标和性能。
[Abstract]:The term "remote sensing" was proposed by E.L.Pruitt and has been developed into a very advanced detection technology since the 1960s. It has been widely used in the field of national economy and national defense, and has produced enormous social and economic benefits. It has become a new method and means for human beings to understand the universe from multi-dimensional and multi-angle. In the process of remote sensing image acquisition, it is inevitable to be affected by various environmental factors, which result in poor visual effect, low resolution and dim brightness, etc. The range of gray scale of remote sensing sensor can not be completely covered by the grayscale of the collected remote sensing image. Therefore, remote sensing image enhancement technology plays an active role in improving image contrast and highlighting some local details. There are two methods for image enhancement: spatial image enhancement and transform domain image enhancement. Spatial image enhancement, as the name implies, is to enhance the image in the spatial domain, in other words, to directly process each pixel in the image. This paper mainly introduces a kind of similar spatial processing method-fuzzy algorithm, and some improved algorithms based on traditional fuzzy algorithm. Transform domain image enhancement method is to transform the original image into a specific transform domain and modify the correlation coefficients in the transform domain to enhance the image accordingly. The methods of image transform from spatial domain to transform domain include Fourier transform, wavelet transform, Contourlet transform, NSCT transform, shearlet transform of shear wave and so on. In this paper, the improved fuzzy algorithm is applied to spatial image enhancement and transform domain image enhancement, respectively. Two algorithms for remote sensing image enhancement are proposed. Firstly, an improved fuzzy algorithm and a spatial image enhancement algorithm are used to enhance the remote sensing image. Because the threshold of the traditional Pal-King algorithm is difficult to determine, this paper uses the improved OTSU algorithm to obtain the threshold from the adaptive algorithm. At the same time, because the fuzzy enhancement is a global fuzzy processing, the effect of detail information enhancement is not obvious. In this chapter, the relative entropy is used as the criterion to enhance the adaptive contrast of domain information. The algorithm enhances the image details on the basis of maintaining the original brightness of the image. Experiments show that the algorithm can obtain good visual effect and more obvious detail information. Second, Shearlet transform is introduced into remote sensing image enhancement for the first time. A remote sensing image enhancement algorithm based on Shearlet transform is proposed. The low frequency coefficients generated by Shearlet transform and the high frequency coefficients of each scale are enhanced by fuzzy contrast and improved fuzzy enhancement respectively. The final enhancement results are obtained by Shearlet inverse transformation. The experimental results show that the proposed algorithm can not only achieve a good visual effect subjectively, but also have a good performance in objective evaluation.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【参考文献】