基于NSCT和Shearlet变换的遥感图像增强研究

发布时间：2018-10-08 09:18

【摘要】：图像,是人类获得信息最直接、最常见的资源。但由于图像的获取与传输过程易受到外界因素的影响,往往使得图像不甚理想,给后续的处理造成一定不便,因此,图像增强在图像预处理阶段是一个关键环节。传统的图像增强主要是在空间域和变换域内处理。模糊算法是一种介于两者之间的方法,即将图像像素映射到模糊域,在模糊域中对相关系数做非线性处理来增强图像的对比度。本文深入系统地研究了基于非下采样轮廓波与剪切波的图像增强算法并结合模糊对比度与多尺度Retinex思想,做出的主要工作如下:1、提出了一种基于NSCT和改进的模糊对比度的图像增强方法。首先对图像进行NSCT变换,其次对低频子带系数采用线性增强的方式来拉伸图像的对比度,用改进的自适应阈值函数来处理高频系数,滤除噪声的影响,最后利用改进的模糊对比度算子来增强图像的边缘和纹理,仿真结果表明,相比于其它传统方法或Pal_King的改进算法,该算法能够明显改善图像的视觉效果并突出图像的细节信息,且抗噪性能良好。2、提出了一种基于Shearlet变换和多尺度Retinex结合的遥感图像增强方法。首先,用多尺度Retinex来调整分解后的剪切波的低频分量,去除照射光对图像的影响,其次,用阈值函数来消除剪切波不同尺度和方向上的高频分量中的噪声。最后,对重构图像进行模糊对比度增强,来提升图像的全局对比度。实验结果表明,本文提出的算法在改善光照条件对遥感图像的影响以及丰富图像纹理细节方面的有效性。
[Abstract]:Image is the most direct and common resource for human to obtain information. However, the process of image acquisition and transmission is easy to be affected by external factors, which often makes the image not ideal, causing some inconvenience to the subsequent processing. Therefore, image enhancement is a key step in the image preprocessing stage. Traditional image enhancement is mainly processed in spatial domain and transform domain. Fuzzy algorithm is a kind of method between them, which maps image pixels to fuzzy domain, and performs nonlinear processing of correlation coefficient in fuzzy domain to enhance the contrast of image. In this paper, the image enhancement algorithm based on non-downsampling contour wave and shear wave is studied in detail and combined with the idea of fuzzy contrast and multi-scale Retinex. The main work is as follows: 1. An image enhancement method based on NSCT and improved fuzzy contrast is proposed. Firstly, the image is transformed by NSCT, then the low frequency subband coefficient is linearly enhanced to stretch the contrast of the image, and the improved adaptive threshold function is used to deal with the high frequency coefficient to filter out the influence of noise. Finally, the improved fuzzy contrast operator is used to enhance the edge and texture of the image. The simulation results show that compared with other traditional methods or Pal_King 's improved algorithm, the algorithm can obviously improve the visual effect of the image and highlight the details of the image. A method of remote sensing image enhancement based on Shearlet transform and multi-scale Retinex is proposed. Firstly, multiscale Retinex is used to adjust the low-frequency component of the shearing wave after decomposition to remove the influence of the irradiated light on the image. Secondly, the threshold function is used to eliminate the noise in the high-frequency components of the shearing wave in different scales and directions. Finally, the global contrast of reconstructed image is enhanced by fuzzy contrast enhancement. Experimental results show that the proposed algorithm is effective in improving the influence of illumination conditions on remote sensing images and enriching the texture details of images.
【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】