森林遥感影像热点的增强算法研究

发布时间：2018-03-04 01:10

本文选题：森林火灾热点　切入点：多光谱红外图像　出处：《电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：通过对森林火灾过程中火势发展规律、遥感图像分析、多光谱红外检测机制的探讨,完成对森林着火初步模型的建立。论文主体以2009年发生在澳大利亚墨尔本市威尔逊国家森林公园的森林大火为原型,借助由中科院地球环境监测中心从中国环境减灾光学卫星A、B上发回的图像热点进行增强分析,精确的增强图像有利于监测系统后期对热点发生火灾及扩散方向的预判。本文侧重研究利用近地光学卫星A、B发回的红外多光谱遥感图像进行分析,总结前人对该类型的数字图像处理方式,引入对森林遥感多光谱图像和遥感红外图像的新式增强算法,具体工作和结论如下:(1)对遥感多光谱图像的主成分进行K-L变换(也称霍特林变换),变换后的各波段图像矩阵只存在自相关项,协方差系数项均为零,将各分量进行图像增强,然后将增强后的分量重新合成,分量间的不相关导致各波段均能较好地保持自己的光谱特性,故合成后的图像能够保留原图像的多光谱特性又达到增强图像目的。通过仿真发现本算法对原图像处理后的光谱保持系数K=0.972046,比直接增强方式对光谱特性的保持要优12.30%,但该处理方式的峰值信噪比、信噪比、信息熵因素均降低。(2)基于平台的加权直方图均衡算法对遥感红外图像进行分析,得到的增强图像与原图和普通直方图均衡图像进行对比,可以从直观看出该图像细节更为清晰。从图像上进行数据分析,可以得到经本文算法对红外图像的处理较普通直方图均衡算法,在图像对比度、峰值信噪比、信噪比、信息熵因素均有所提高。峰值信噪比等因素的提高说明本算法处理后的图像更贴近实际火灾热点分布图。经过测算本算法热点区域对比度相对普通直方图均衡算法增加约为47.22%,。而对比度的提高,有利于准确判断火灾热点。
[Abstract]:The development law of fire intensity, the analysis of remote sensing image and the mechanism of multi-spectral infrared detection in forest fire are discussed. The main body of the paper is based on the forest fire that occurred in Wilson National Forest Park in Merben, Australia on 2009. With the help of the hot spots of the images sent back from the China Environmental Mitigation Optical Satellite AHB by the Earth Environment Monitoring Center of the Chinese Academy of Sciences, Accurate enhanced images are helpful to predict the direction of fire and diffusion of hot spots in the later stage of the monitoring system. This paper focuses on the analysis of infrared multispectral remote sensing images from the near Earth Optical Satellite (AMAB). This paper summarizes the previous digital image processing methods and introduces a new enhancement algorithm for forest remote sensing multispectral images and remote sensing infrared images. The specific work and conclusions are as follows: (1) K-L transform (also known as Hotlin transform) for the principal components of multispectral remote sensing images. After the transformation, only autocorrelation items exist in each band image matrix, the covariance coefficients are all zero, and each component is enhanced. Then the enhanced components are recombined, and the uncorrelated components result in each band being able to maintain its own spectral characteristics. Therefore, the synthesized image can preserve the multi-spectral characteristics of the original image and achieve the purpose of enhancing the image. Through simulation, it is found that the spectral holding coefficient K _ (0.972046) of the original image processed by this algorithm is better than that of the direct enhancement method. Excellent 12.30, but the peak signal-to-noise ratio of the processing, The signal-to-noise ratio (SNR) and information entropy are all reduced. (2) based on the weighted histogram equalization algorithm based on the platform, the enhanced image is compared with the original image and the ordinary histogram equalization image, and the enhanced image is compared with the original image and the ordinary histogram equalization image. By analyzing the data on the image, we can get that the infrared image is processed by this algorithm compared with the ordinary histogram equalization algorithm, and the image contrast, the peak signal to noise ratio, the signal-to-noise ratio, and the signal to noise ratio can be obtained. The increase of peak signal to noise ratio (PSNR) shows that the image processed by this algorithm is closer to the actual fire hot spot distribution map. The contrast of the hot spot area of this algorithm is compared with the ordinary histogram equalization calculation. The increase of the method is about 47.221.The contrast is improved. It is propitious to judge the hot spot of fire accurately.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【参考文献】