基于变换编码和分布式编码的谱带排序优化的卫星图像通信方法

发布时间：2018-02-14 14:30

本文关键词： 有损压缩变换编码速率控制多谱带排序遥感图像通信分布式编码　出处：《哈尔滨工业大学》2017年博士论文　论文类型：学位论文

【摘要】：目前,遥感中心提供越来越多的地球表面可用卫星图像。一方面,卫星图像提供大量信息,每一张图像大小在几十兆字节到千兆字节范围之间,图像压缩可以有效的减少存储的需求。另一方面,目前,全球定位系统提供用户的位置和信息。因此,使用卫星导航已成为生活中不可或缺的一部分,同时,传输这些卫星图像成了一个巨大的挑战。此外,高光谱图像已经在独立的陆地地图服务,以及政府和军事活动等各种各样遥感项目得到了广泛应用。考虑到高光谱卫星图像中的一些频带受到卫星和地面站之间的大气以及信道噪声的干扰,我们将同一个高光谱图像的不同版本进行联合处理,以获得更高的图像质量。因此,研究高光谱卫星图像通信方案以提高接收图像的质量成为了新的研究热点。本文提供了卫星图像通讯的有效方法。在通信系统中,由两部分组成;压缩和传输。本文的第一部分介绍了两个卫星使用小波变换和速率控制卫星图像的图像压缩方法。第一个压缩方法是基于从原始卫星图像众多相关联的频带中去除一些小波子带。被分解的子带由使用带之间的相关系数决定,然后重建解码器从最相关的子带到那些被删除的子带。这种方法减少了高质量卫星图像的储存大小。来自ETM的多频谱图像。实验结果表明,该方法将平均多光谱图像质量提高了3-11d B。第二个压缩方法介绍了一个立足于码率控制的卫星图像有损压缩的简单方法,该方法的目标是减少植物特征的影响。这个压缩技术是基于一个新的速率控制方案的植被组(即。组包含最重要的乐队,植被特征提取)编码比特率高于其余组。该研究是在几个卫星图像上进行的,来自ETM的多频谱图像;通过飞船可见分光仪采集的多频谱遥感图像。在所有实验结果中,传统的压缩方法对卫星图像的植被特征表现出更大的影响。此外,所提出的方法对植被特征影响较小。这些结果验证了所提方法的有效性。第二部分介绍了两个无线通信构架,卫星传感图像问题中解决大量接收端情况下的无线编码和基于Soft Cast和分布式编码。本文将这些框架进行了扩展,用来解决多波段的遥感图像问题。第一个通讯方法,称为Hyper Cast,使用一个集群转换的方法。这种方法将高光谱图像波段分成组,然后采用变换从每组删除冗余信息。随后,传输能量将根据他们的能量分布直接分配到量化数据上,无需FEC的参与。Hyper Cast分别在Line Cast,Soft Cast-3D和传统框架上广播PSNR增益高达6.98d B,3.48dB和6.14d B。第二种通讯方法是基于DWT和分布式信源编码实现的。小波变换之前,高光谱的预处理是进行两步波段订购和规范化。对DWT操作,我们进行一个三级划分,每个频谱图像划分为10个小波子带,其中9个子带是图像的细节描述。最后的一个级别(LL-LL-LL)为图像的近似描述。对于LL-LL-LL子带,我们采用了基于信源分布式编码的培基编码技术来提高压缩效率,同时降低编码复杂度。小波分解后细节和陪集值传播类似于前面的方案没有FEC。广播中的实验结果表明,该方案分别比Line Cast,Soft Cast-3D和常规框架提高了6.91d B,3.00dB和5.63d B平均图像质量。此外,所提出的方案比所有最先进的方案具有较少的编码时间。最后,此通信方案提出了高效的排序算法来提高压缩性能。与平均PSNR上没有频带排序高达1.30d B的情况相比,所提出的频带排序算法具有更好的性能。
[Abstract]:At present, the remote sensing center provides satellite images of the earth's surface can be more and more. On the one hand, the satellite images provide a lot of information, each image in size between tens of megabytes to gigabytes, image compression can effectively reduce the storage requirements. On the other hand, at present, the global positioning system to provide user location and information therefore, the use of satellite navigation has become an integral part of life, at the same time, the transmission of satellite images has become a huge challenge. In addition, the hyperspectral image has been in the land map service, and government and military activities of various remote sensing projects have been widely used. Considering some hyperspectral bands satellite images are between the satellite and the ground station and the atmospheric channel noise, we combine the different versions of the same hyperspectral images were combined to obtain treatment. Get better image quality. Therefore, the communication scheme of hyperspectral satellite images to improve the quality of the received image has become a new hotspot. This paper provides an effective method of satellite image communication. In the communication system, is composed of two parts; compression and transmission. The first part of this paper introduces two satellite images wavelet transform and rate control of satellite image compression methods. The first method is compressed from the original satellite images of many associated bands in the removal of some wavelet subband decomposed subband. Based on the correlation coefficient between the belt by using the decision and reconstruction from the most relevant sub decoder to those deleted sub bands. This method reduces the size of high quality satellite image storage. Multi spectral images from ETM. The experimental results show that this method will improve the quality of the average multi spectral image compression 3-11d B. second The method introduces a simple method based on satellite image compression rate control method, the goal is to reduce the influence of plant characteristics. The compression technology is a new vegetation group rate control scheme based on (i.e. group contains the most important bands, vegetation feature extraction) the encoding bit rate is higher than the rest group. The study was performed in several satellite images, multi spectral images from ETM; multi spectral remote sensing image through visible spectrometers. The spacecraft in all of the experimental results, traditional compression methods of satellite images of vegetation characteristics showed a greater impact. In addition, the proposed method has little effect on vegetation features. These results validate the effectiveness of the proposed method. The second part introduces the two wireless communication architecture, wireless receiver case encoding a satellite sensing image problem and based on Soft Cast and distributed encoding. In this paper, the framework is extended to solve the problem of multi band remote sensing images. The first communication method, called Hyper Cast, using a cluster conversion method. This method of hyperspectral image band is divided into groups, and then the transformation from each group to delete the redundant information. Then, the transmission of energy according to the energy distribution of their allocated directly to the quantitative data, without the involvement of the FEC.Hyper Cast Cast Soft Cast-3D in Line, and the traditional framework of broadcast PSNR gain up to 6.98d B, 3.48dB 6.14d and B. second kinds of communication method is based on DWT and distributed source encoding implementation. Wavelet transform before pretreatment the two step is the high spectral band ordering and standardization. For DWT operation, we performed a three grade classification, each spectrum image is divided into 10 wavelet sub bands, including 9 sub bands is the details of the image. The last. One level (LL-LL-LL) description for the approximation of the image. For the LL-LL-LL subband, we used to improve the compression efficiency based on distributed source encoding technology medium encoding, and reduce encoding complexity. After wavelet decomposition details and coset similar to the value propagation scheme without FEC. radio experimental results this scheme, respectively than the Line Cast, Soft Cast-3D and 6.91d B to improve the conventional framework, the average 3.00dB and 5.63d B image quality. In addition, the proposed scheme is less than all of the most advanced scheme of encoding time. Finally, the communication scheme is proposed to improve the algorithm efficient compression performance. Compared with no band sort of up to 1.30d B PSNR on average, band sorting algorithm proposed has better performance.

【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TN927.2

【相似文献】