Research On Time Sequence Compression Algorithms Based On Time Domain Processing

The data compression technology with high compression ratio and high SNR isone of targets in data processing. The classical statistic coding methods, such asHuffman Coding and Arithmetic Coding, can be applied to most types of datatheoretically, but the higher compression ratio can not be guaranteed for any kinds ofdata. So far, much effort has been made for the data compression methods to deal withdata sources that have various probability distributions.A new data compression algorithm which can be widely used in compression ofgeneral time domain sequences is introduced in the thesis. In this algorithm, theADPCM data compression algorithm is first utilized to compress the input originalsequence, then the residual produced by ADPCM coding of the original sequence isencoded by the optimized scalar quantization making use of the Lloyd-Max algorithm.Since the probability distributions of the codes resulting from the original sequencecoding and its residual coding are not uniform, the entropy coding, such as Huffmancoding, is applied to the codes of ADPCM and Lloyd-Max encoding, so as to get morecompression. In order to reduce the big values occur in the residual, special treatmentmethods are designed and implemented. For the original time domain sequencessampled with accuracy of16bits/sample, three coding bit rates, i.e.9.00bits/sample,8.00bits/sample and7.25bits/sample, are designed and supported in the algorithm.Test and evaluation have shown that the performances of designed algorithm,including compression ratio and SNR, are good for the time domain sequences that havedifferent probability distribution, and the calculation complexity of the algorithm islower enough for hardware implementation, therefore the algorithm can be generallyused in many applications. Furthermore, this algorithm can be easily adopted forlossless data compression.