基于虚警概率最大准则的小波阈值去噪算法研究

本文分析了随机噪声的小波变换系数在不同尺度上的传递特性和噪声信号奇异性与小波模极大值的关系,阐述了基于阈值的正交小波变换去噪法,在此基础上提出了一种基于能量元和Neyman-pearson准则的小波阈值去噪算法。其小波阈值的选取是依据设定虚警概率后得到的检测概率最大的准则,并且阈值是随各尺度独立的。通过仿真实验,验证了该算法的有效性和优异性。

Research on Algorithm of Wavelet Threshold Value De-noising Based on Maximum Criterion of False Alarm Probability

According to the characteristics of random noise wavelet transform on the different scale and the relationship between of noise Lipschitz and its wavelet transform modulus maxima, the orthogonal wavelet threshold de-noising method is explained, and then a de-noising method based on energy-member and Neyman-Pearson Criterion is proposed. The selection of its threshold is according to the criterion that the detection probability is maximized in given false alarm rate probability, and it is independent on each scale. Numerical simulations show that this de-noising method is effective and excellent.