【摘要】 基于平稳离散小波变换 (SDWT)研究了一种改进阈值消噪方法 :给出了一种新的跨尺度阈值计算方法 ;使用将软、硬阈值估计的细节系数进行加权平均的细节系数估计方法 ;通过补充大尺度细节系数降低消噪误差的方法。与经典的软、硬阈值方法比较 ,大量的仿真试验说明本文的方法在信号畸变、误差、光滑度等性能指标之间取得了合适的折衷。将此方法应用于动态触觉传感器检出信号的消噪 ,提高了检出信号的信噪比 ,显著减小了误差和信号畸变 ,取得了显著的性能改进更多还原

【Abstract】 Based on stationary discrete wavelet transform (SDWT), de-noising by enhanced thresholding method was investigated. New formulas were given on the thresholds of different scales, the estimated detail coefficients, and the detail coefficients complementing on the largest scale. Compared to the classical soft and hard thresholding method, a large amount of computer simulations prove this method gets a fine tradeoff among signal mutation, error and smoothness. This method was applied to de-noise the sampled signal of a PVDF dynamic tactile sensor. And the experimental results prove the increase of SNR, the decrease of signal error and mutation, and the remarkable better performance.更多还原