论文部分内容阅读
本文提出一种用于计算机辅助诊断前处理的医学信号去噪新方法。方法：１．对信号进行ＤＷＴ。２．在ｓ＝２１上，利用邻近尺度ＷＴ系数的空间相关性进行选择性去噪。３．在ｓ≥２２上，运用阈值法（Ｓｏｆｔ，ＨａｒｄＴｈｒｅｓｈｏｌｄｉｎｇ）去噪。４．信号重构。结果：分别采用本文方法及传统方法对突变信号进行去噪与结果比较。结论：本文方法在获得信噪比增益的同时，能够保持对信号突变信息的良好分辨且处理速度快、便于在微机上实现。
This paper presents a new method of medical signal denoising for computer-aided diagnosis and pre-processing. Methods: 1. DWT the signal. 2. At s = 21, selective denoising is performed using the spatial correlation of WT coefficients of neighboring scales. 3. In s ≥ 22, using the threshold method (Soft, HardThresholding) denoising. 4. Signal reconstruction. Results: The proposed method and the traditional method were used to denoise the mutation signal and compare the results. Conclusion: This method can gain good signal-to-noise ratio resolution and fast processing speed while gaining signal-to-noise ratio gain, and is easy to implement on a microcomputer.