摘要
遥感大数据处理是典型的大数据应用方向之一.由于遥感图像时空分辨率高、数据量大,相邻像元间存在较大的冗余信息,这给数据传输和数据存储带来一定的困难.为此,本文提出一种基于奇异值分解(SVD)的遥感图像压缩方法,首先通过SVD分解获得压缩矩阵,然后采用局部压缩矩阵生成原始图像的近似图像.在分解过程中,通过压缩矩阵的奇异值曲线获取奇异值的衰减曲线以及对图像的贡献程度.选取不同的压缩值可以实现遥感图像的压缩.实验表明,基于奇异值分解的遥感图像压缩方法能够有效地实现遥感图像的压缩,具有良好的压缩效果.
Big data processing of remote sensing image is a typical application in the big data processing field.Due to the high temporal-spatial resolution and data volume of remote sensing image,there exists large redundancy between adjacent pixels,which poses some difficulties to data transmission and data storage.To resolve this problem,a remote sensing image compression method based on singular value decomposition(SVD)was proposed.Firstly,the compression matrix was obtained by SVD decomposition,and then the local compression matrix was used to generate an approximate image of the original image.In the decomposition process,one could obtain the attenuation curve of the singular values and their contribution to the remote-sensing image from the singular value curve of the compressing matrix.Finally,we could realize the remote-sensing image compression by selecting different compression values.The experiment results show that the proposed method can effectively realize the remote-sensing image,hence a good compression effect.
作者
陈文青
范恩
胡波
CHEN Wenqing;FAN En;HU Bo(School of Information Engineering,Shaoxing Vocational&Technical College,Shaoxing,Zhejianig 312000;Department of Computer Science and Technology,Shaoxing University,Shaoxing,Zhejiang 312000;Ningbo Meteorological Administration,Ningbo,Zhejiang 315012)
出处
《绍兴文理学院学报》
2020年第10期109-113,共5页
Journal of Shaoxing University
基金
浙江省教育厅一般科研项目“基于分布式平台Spark的大数据挖掘算法优化研究”(Y201636830)
浙江省科技厅公益工业项目“基于大数据理论的燃煤机组性能优化关键技术研究与应用”(LGG18F020012)
宁波市自然基金项目“基于深度学习的数值预报产品降尺度研究——以宁波市为例”(2019A610455)
宁波市科技局公益性项目“基于深度学习的宁波大雾短时预报预警技术研究”(2019C50001)
国家自然科学基金项目“基于中智集的多传感器空域目标抗差航迹关联与航迹融合”(61703280).
关键词
遥感图像数据处理
奇异值分解
图像压缩
remote sensing image data processing
singular value decomposition(SVD)
image compression
