基于PCD算法的信号恢复重构研究

利用μ=1和线搜索求μ方法,该文首先研究了利用分离替代函数算法(SSF)和平行坐标下降算法(PCD)求解无约束优化问题的性能,实验表明PCD算法性能优于SSF算法;其次,研究了模糊噪声下的参数λ与ISNR的变化关系;最后,研究PCD算法分别对σ2=2和σ2=8二维模糊噪声图像信号的重构。实验表明PCD算法有良好的去模糊消噪能力,为图像检测识别之前提供必要的恢复重构条件。

Research on signal reconstruction based on the algorithm of parallel coordinate descent

Using μ=1 and using line search to solve μ, the paper first studies that using Separable Surrogate Functional algorithm ( SSF ) and Parallel Coordinate Descent ( PCD ) algorithm solves uncon-strained optimization problems in order to study the performance of different algorithms. The experimental results show that the performance of PCD algorithm is better than that of SSF algorithm. Secondly, the paper studies the variable relationship of parameter λ and ISNR under the blur and noise. Finally, the research on two dimensional images of blur and noise based on σ2 =2 and σ2 =8 uses PCD algorithm to reconstruct original image, respectively. The experimental results show that the PCD algorithm has good performance to deblur and denoise. Before the detection and identification of image, this process provides the necessary reconstruction method.