CAJ | 학술논문

在求解基追踪问题的线性化Bregman迭代方法基础上,结合了广义逆的迭代技术得到一种稀疏信号重构的新算法。该算法在计算Moore-Penrose广义逆时,采用了迭代计算的方式,与算法本身相结合使得仅有矩阵向量乘积运算,避免了奇异值分解的较大工作量。通过数值试验可知,新算法相对线性化Bregman算法在计算时间上约减少了2/3,同时信号的恢复效果也是稳定有效的。因此,新算法是一种有效可行的信号重建算法。
재구해기추종문제적선성화Bregman질대방법기출상,결합료엄의역적질대기술득도일충희소신호중구적신산법。해산법재계산Moore-Penrose엄의역시,채용료질대계산적방식,여산법본신상결합사득부유구진향량승적운산,피면료기이치분해적교대공작량。통과수치시험가지,신산법상대선성화Bregman산법재계산시간상약감소료2/3,동시신호적회복효과야시은정유효적。인차,신산법시일충유효가행적신호중건산법。
Combined with the iterative method of generalized inverse, a new algorithm for sparse signal re-construction is developed based on linearized Bregman iteration for solving the basis pursuit problems. Dur-ing calculating Moore-Penrose generalized inverse, the algorithm only has the matrix-vector multiplication by iteration combined with its own,so that the singular value decomposition( SVD) is avoided. The numer-ical experiments show the computation time has reduced by about 2/3 than that of original algorithms. Meanwhile,the recovery of signals is stabe and effective. So this new algorithm is a feasible signal recon-struction algorithm.