论文部分内容阅读
针对分数低阶α稳定分布信号具有显著脉冲这一特性,参照最优小波包基选择的熵准则,提出了一种针对该类噪声的最优小波基选择新准则,解决了以往最优小波基选择的熵准则中,由于需计算二阶统计量而不适合分析分数低阶α稳定分布信号的问题。还得出了针对该类信号,在运用该新准则进行判别最优小波基时,只需对各个小波基函数在第1层尺度下的变换系数进行对比判别的结论,极大地减化了以往多尺度的判别过程。将新准则应用于医学超声图像消噪时,获得了较好的效果。
Aiming at the characteristic that the fractional low-order α stable distribution signal has significant impulses, a new criterion for the selection of optimal wavelet bases for such noise is proposed by referring to the entropy criterion of optimal wavelet packet basis selection, In the selected entropy criterion, it is unsuitable for the problem of analyzing fractional lower-order α-stable distribution signals due to the need of calculating second-order statistics. It also concludes that for the class of signals, when using this new criterion to discriminate the optimal wavelet base, it is only necessary to compare the transform coefficients of each wavelet basis function at the first layer scale, which greatly reduces the previous Multi-scale discrimination process. When the new criterion is applied to the medical ultrasound image denoising, the better effect is obtained.