摘要
VDMOS器件中的1/f噪声是表征该器件质量和可靠性的重要敏感参数,在检测器件噪声过程中,1/f噪声是被淹没在大量白噪声背景下的一种微弱信号,为了能准确检测出1/f信号,本文提出了基于小波熵软阈值1/f检测方法。利用1/f噪声及白噪声在不同分解层次上小波系数熵的变化特点,根据信号子带高频小波系数的最大小波熵是由白噪声引起的,对不同分解尺度上的高频系数进行小波熵阈值量化处理,最后对信号进行重构,去掉了严重的白噪声,实现1/f信号的检测。实验表明与传统软阈值去噪的性能分析比较可知,该方法提高了检测1/f信号的准确率。
The 1/f noise in VDMOS device is an important sensitive parameter for characterizing the quality and reliability of the device. In the process of device noise detection, 1/f noise is a weak signal submerged in the context of a large number of white noise. In order to accurately detect 1/f signal, a wavelet entropy based soft thresholding method is proposed in this paper. Considering different characteristics of wavelet coefficient entropies for 1/f noise and white noise changes at different levels, and noticing the fact that the maximum wavelet entropy for the high frequency wavelet coefficients of the sub-band signal is caused by white noise, the wavelet entropy threshold quantization process is conducted on high frequency coefficients at different scales. Finally the signal is reconstructed to remove the severe white noise and detect the 1/f signal. Experiments show that compared with the performance of traditional soft thresholding methods, the method improves the accuracy of the signal detection.
出处
《电测与仪表》
北大核心
2013年第11期42-46,共5页
Electrical Measurement & Instrumentation
基金
国家自然科学基金资助项目(61271115)
