【摘要】 传统的离散小波变换(Discrete Wavelet Transform,DWT)卷积算法通常采用延拓方法来解决边界效应,处理速度会受到一定影响。为了进一步提高使用DWT的实时性,提出一种新的离散卷积快速实现算法。该方法利用前后端卷积结果对边界小波系数进行校正,有效地解决了卷积运算中的边界效应问题,并减少了计算量。仿真试验和工程应用结果表明该算法能用于分析平稳信号和非平稳信号,具有很高的分解和重构精度以及很好的边界处理性能,并且计算速度比传统方法有较大提高,具有显著的工程应用价值。更多还原

【Abstract】 The traditional convolution algorithm for discrete wavelet transformation( DWT) usually makes use of extension to solve the boundary effect,thus the processing speed is affected. To further improve the real-time performance of DWT,a novel fast algorithm for implementing discrete convolution was proposed. With this approach,the convolution results at both ends were used to correct the boundary wavelet coefficients,so the boundary effect of convolution operation was effectively solved and the computational complexity was reduced. Simulation test and engineering application results showed that the proposed algorithm is suitable for analyzing stationary signals and nonstationary signals; it has a high decomposition / reconstruction accuracy and a good boundary processing performance; moreover,its calculation speed is much faster than that of the traditional method,hence it is valuable for engineering application.更多还原