摘要
针对信号在过完备库上分解中原子生成速度慢的难题,提出了一种原子生成的快速算法.首先根据原子的尺度把原子分成小原子和大原子2类.对于小原子,因为其能量集中在较小的范围,所以用小范围生成的局部原子代替整个原子.对于大原子,先生成相应的较小原子,然后通过插值方法生成大原子.实验结果表明,当信号长度为256时,本算法在重建信号的质量没有任何改变的条件下,原子生成的速度比传统算法提高了4. 7倍.
It is one of the main problems in signal decomposition in over-complete dictionary that the atom construction process is very slow. To solve this problem, a new fast algorithm was proposed. In the algorithm, all atoms are divided into two categories： small and large atoms, according to their scales. Because the energy of a small atom concentrates in a small region of the whole atom, it is constructed within the region of energy concentration, and the whole atom is represented by the locally constructed atom. A large atom is constructed by interpolation after a corresponding small atom has been constructed. Experimental results show that, when the length of the signal is 256, the proposed algorithm is 4.7 times faster than traditional methods with the same signal quality.
出处
《西南交通大学学报》
EI
CSCD
北大核心
2005年第3期402-405,共4页
Journal of Southwest Jiaotong University
基金
国家留学基金资助项目(21851039)
四川省应用基础研究项目 (03JY029 048 2
04JY029 059 2)
教育部留学回国人员科研启动基金资助项目(教外司[2004]527号)
关键词
信号处理
稀疏分解
过完备原子库
快速算法
signal processing
sparse decomposition
over-complete dictionary of atoms
fast algorithm
