基于MPSoC平台小波变换并行算法研究

发布时间：2018-02-22 09:25

本文关键词： 信号处理快速傅里叶变换小波变换多核嵌入式系统并行计算　出处：《计算机工程与应用》2017年04期 　论文类型：期刊论文

【摘要】：快速小波变换是数字信号处理面临的一个重要问题,针对并行小波算法展开研究,缩减小波变换中卷积运算的规模,提高小波变换过程中的并行效能,以实现小波变换的快速并行计算。通过FFT矩阵代入计算,消去了并行计算过程中的同步通信,降低了乘法运算次数。对算法思想进行了理论分析,说明新算法在短小数据分段情况下能够减少50%~75%的乘法操作;通过搭建两种不同平台进行了对比测试,证明了算法的先进性与有效性。基于FFT矩阵的并行小波变换算法是一种稳定有效的经典小波并行算法。
[Abstract]:Fast wavelet transform is an important problem in digital signal processing. Research on parallel wavelet algorithm is carried out to reduce the scale of convolution operation in wavelet transform and to improve the parallel efficiency in the process of wavelet transform. In order to realize the fast parallel computation of wavelet transform, the synchronous communication in the parallel computing process is eliminated and the multiplicative operation times are reduced through the FFT matrix substitution calculation. It shows that the new algorithm can reduce the multiplication operation by 50% and 75% in the case of short data segmentation. It is proved that the algorithm is advanced and effective. The parallel wavelet transform algorithm based on FFT matrix is a stable and effective classical wavelet parallel algorithm.
【作者单位】：中南大学信息科学与工程学院;广州大学计算机科学与教育软件学院;
【基金】：国家自然科学基金(No.61073037,No.61272496,No.61272151) 教育部博士点基金资助项目(No.20110162110043)
【分类号】：TN911.7

【相似文献】