CAJ | 학술논문

针对匹配追踪算法（ MP）在信号去噪中存在计算量大的缺点，将基于多目标的粒子群优化算法应用到MP算法去噪中。提出利用两个目标函数来优化算法，从而减少计算量。目标函数分别取信号或信号残差与原子两者的相关系数绝对值的倒数和信号残差。仿真结果表明：改进的算法与基础MP算法相比，明显地去除了尖峰噪声，而且在信噪比和均方差上有明显的改善。
침대필배추종산법（ MP）재신호거조중존재계산량대적결점，장기우다목표적입자군우화산법응용도MP산법거조중。제출이용량개목표함수래우화산법，종이감소계산량。목표함수분별취신호혹신호잔차여원자량자적상관계수절대치적도수화신호잔차。방진결과표명：개진적산법여기출MP산법상비，명현지거제료첨봉조성，이차재신조비화균방차상유명현적개선。
Considering large amount of calculation of the MP( matching pursuit) algorithm in the signal de-noisng, the multi-objective-based particle swarm optimization( PSO) algorithm was applied to the MP de-noi-sing;and making use of two objective functions to optimize this method and to reduce the amount of calculation was proposed, in which, the objective function adopts the reciprocal and signal’ s residual of the correlation coefficient’ s absolute value between the signal and the atom or the signal residual and the atom.The simula-tion results show that compared with the MP algorithm, the proposed de-noising algorithm can get rid of the spike noise obviously together with significant improvements on noise ratio and the mean square error.