用SAVQ技术实现太阳射电频谱的数据压缩

对自适应矢量量化（SAVQ）技术在太阳射电频谱数据压缩中的应用进行了有意义的讨论．给出一种压缩比和失真可调的SAVQ压缩方法，指出提高数据压缩比的有效方法之一为样本集成，同时介绍了用该技术对北京天文台太阳射电频谱仪上的两个波段（1．0GHz－2．0GHz和2．6GHz－3．8GHz）的观测数据进行压缩的实用效果，并与国际同行所用的ICON方式进行性能比较．最后展示了数据压缩技术及其在天文上的应用．

COMPRESSION OF THE DATA OF SOLAR RADIO SPECTROMETER WITH SAVQ

In this paper, we discuss the application and significance of Self-Adaptive Vector Quantization (SAVQ) technology in the data compression of the solar radio spectrometer. A compression method of SAVQ, in which both the compression ratio and the distortion may vary, is presented. Furthermore, it is pointed out that the integrated sample is one of the valid methods in which the data compression ratio is raised. At the same time, we introduce the practical efficiency with which the data of Beijing Observatory's solar radio spectrometer at 1.0GHz-2.0GHz and 2.6GHz-3.8GGz are compressed and the result is compared with that of ICON in Switzerland.Finally, the prospects of the data compression technique and its application in astronomy are predicted.The paper consists of five parts. In the first one, the significance of the lossy data compression of the solar radio spectrometer is discussed. In the second part, the self-adaptive vector quantization technology and its characteristics in astronomical data compression are introduced. Meanwhile, the heterogeneous quantization method in power is put forward according to the data characteristics. In the third part, it is pointed out that the compression ratio can be raised with the integrated sample. A large sample can provide more accurate statistical message and reduce the proportion with which the compression algorithm occupies memory in the total data. In the fourth part we test some samples of the solar radio spectra of Beijing Observatory.The result is that the compression ratio is larger than 260 and the full distortion is merely less than 25% of the maximum sample. In the last part, the various merits of the application of the data compression technique in astronomy are described in details.