基于预测的高光谱图像无损压缩技术研究

发布时间：2018-07-12 08:15

本文选题：高光谱图像无损压缩 + 边缘检测　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：高光谱图像是一种空间和谱间分辨率都很高的三维立体遥感图像。随着成像光谱技术的发展，高光谱图像的数据量急剧增加，给高光谱数据的传输和存储带来了巨大的压力，严重制约了高光谱遥感的发展。因此研究高光谱图像压缩算法对高光谱图像的广泛应用有着重大的意义。本文首先介绍了高光谱图像的特点及应用，，分析了高光谱图像无损压缩技术的研究现状。然后描述了无损压缩的基本理论，特别说明了无损预测压缩的基本框架及现有技术。最后在现有的高光谱图像压缩算法基础上，主要做了两方面的工作。一方面是提出了基于边缘检测的改进SLSQ高光谱图像无损压缩方法。与SLSQ相比，该方法在已有的谱内预测模式和谱间预测模式上增加了无预测模式，在谱内中值预测的基础上增加了对斜边缘的检测，在谱间预测中提出了自适应边缘预测。测试结果表明，对AVIRIS1997高光谱图像和CCSDS测试图像，该方法得到的压缩比比SLSQ分别提高了0.04和0.07。另一方面是对CCSDS多/高光谱图像无损压缩标准123.0-B-1（简称为MHDC标准）进行了性能分析。该标准中共包括12个用户自定义参数，通过对这些参数的测试及分析得到了一组优化参数，最后对标准性能进行了评测。测试结果表明，优化参数下的压缩性能明显优于参考参数，MHDC标准对AVIRIS1997高光谱图像和CCSDS测试图像性能居中。
[Abstract]:Hyperspectral image is a three-dimensional remote sensing image with high spatial and interspectral resolution. With the development of imaging spectrum technology, the data volume of hyperspectral images increases dramatically, which brings great pressure to the transmission and storage of hyperspectral data, which seriously restricts the development of hyperspectral remote sensing. Therefore, the study of hyperspectral image compression algorithm is of great significance to the wide application of hyperspectral image. In this paper, the characteristics and applications of hyperspectral image are introduced, and the research status of lossless compression of hyperspectral image is analyzed. Then the basic theory of lossless compression is described, especially the basic frame of lossless prediction compression and the existing techniques. Finally, based on the existing hyperspectral image compression algorithm, two aspects of work are mainly done. On the one hand, an improved SLSQ hyperspectral image lossless compression method based on edge detection is proposed. Compared with SLSQ, this method adds no prediction model to the existing prediction model in spectrum and interspectral prediction model. On the basis of median prediction in spectrum, the detection of oblique edge is added, and adaptive edge prediction is proposed in inter-spectral prediction. The results show that the compression ratio of AVIRIS1997 hyperspectral image and CCSDS image is increased by 0.04 and 0.07 than SLSQ, respectively. On the other hand, the performance of CCSDS multispectral / hyperspectral image lossless compression standard 123.0-B-1 (referred to as MHDC standard) is analyzed. The standard includes 12 user-defined parameters. A set of optimized parameters is obtained by testing and analyzing these parameters. Finally, the standard performance is evaluated. The test results show that the compression performance of the optimized parameters is better than that of the reference parameter MHDC standard for AVIRIS1997 hyperspectral images and CCSDS test images.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【参考文献】