基于隐马尔科夫树模型的小波域压缩采样信号重构方法.pdfVIP

第 24 替第4 期 电子测量与仪器学报 Vol.24 NO.4 314 • JOURNAL OF ELECTRONIC MEASUREMENTAND INSTRUMENT 2010 年4 月 DOI: 1O.3724/SP.J.l 187.20l0.ω314 基于隐马尔科夫树模型的小波域压缩 采样信号重构方法* 赵贻执J..)宇平 戴志坚 (电子科技大学自动化工程学院，成都611731) 摘 要:压缩传感理论利用信号的稀疏性，对其非自适应线性投影进行Æ缩采样，通过最优化问题准确建构原始信号。 传统囊构算法仅利用了恼号的稀疏性，而未对转换盾的倍号结构进行分析。提出了一种摹于4 状态的隐马尔科夫树模服的小 披域压缩采样信号的重构方法，相对 2 状态的隐马尔科夫树模型，该模型能够挟取相邻尺度小被系数的更多相关特性，通过 仿真结果表明，诙算法具有更高的重构精度。 提键词:压缩采样;非自适应钱性投影;小披变换;隐马尔科夫树模现 中剧分赞号: TN91 1.7 )c献标识码:A 阔家标准学科分提代码: 510.40 Compressive sampling signal reconstruction in wavelet-domain based on hidden Markov tree model Zhao Yijiu Wang H叫un Dai Zhijian (School ofAuωmation Engine明ing， University of Electronic Science and Technology of China, Chengdu 611731 , China) Abstract: Compressed sensíng theory enables the reconstruction of sparse signal from a small number of non崛adaptives linear projections. Conventional reconstructíon algorithm involves linear programming or greedy algo- rithms , these reconstruction techniques are generic and 刷刷me no particular struc阳re in 伽signal aside from sparsity. The compressive sampling signal reconstructíon in wavelet domain is inspired based on tow唰state wavelet hidden Markov 衍ee model. In也is paper, we propose a four-state wavelet Hidden Markov Tree model , it can capωre more in- terscales dependencies of wavelet coefficíents between two neighboring scales,the simulation shows that it recoDstruc- tiOD precísioD is ímproved. Keywords: compressive sampling; Don蝴adaptives linear projection; wavelet transform; hidden markov tree model 同 古 稀疏性为基础对倍号进行重构，来对倍号的结掏进 1 引|