【摘要】 信号的稀疏表示在信号处理的许多方面都有重要的应用,但稀疏分解计算量十分巨大,难以被推广而实现产业化。模拟退火算法模拟高温金属降温的热力学过程,广泛应用于求解组合优化问题。本文将模拟退火算法运用到信号的稀疏分解中,首先随机产生一组原子参数组,然后分别计算每个原子与信号或信号残差的内积的绝对值,找出内积绝对值最大的原子参数组并对它进行模拟退火处理,用处理的结果作为匹配追踪(MP)过程中每一步的最优解。在计算过程中利用原子的特性,进一步提高了信号稀疏分解的速度,并用本文提出的算法与其它方法进行了比较,实验结果表明了该算法的有效性。更多还原

【Abstract】 Sparse representation of signals has found many important applications in signal processing,but the computational burden in the signal sparse decomposition process is so huge that it is almost impossible to apply it to industrialization.The Simulated Annealing(SA) algorithm simulates the thermodynamics process of metal heat temperature-decreasing,and it has been widely used in solving many problems of combination optimization.In this paper,the SA algorithm is applied to signal spare decomposition.A group of random parameter sets are generated at first,and the absolute values of inner products of atoms with these parameters sets with the signal or the residual of the signal are calculated.The set of parameters corresponding to the biggest value are processed with SA.The parameters set finally got by SA is regarded as that of the optimal atom at each step of Matching Pursuit(MP).Utilizing the property of atoms,the speed of signal sparse decomposition is raised a lot.The proposed algorithm is compared with others in the paper,and finally experimental results show the effectiveness of the proposed algorithm.更多还原