在信号去噪声中 ,要同时去除脉冲噪声和白噪声 ,并保持信号的突变现象 ,有一定的困难。研究了左预测和右预测 ,及交叉有效 (CV)平滑法 ,并证明了 CV平滑中的两个限制条件。提出了有限预测混合 CV平滑法去噪 ,能同时达到以上目的。分析了该方法的根信号特征。并通过模拟试验检验了该方法 ,并与有限混合中位数滤波 (FMH )法和自适应小波收缩法 (WSA )作了比较 ,试验结果说明了该方法是切实可行的 ,并优于 FMH和 WSA。 In the signal de-noise, to remove both impulse noise and white noise, and to maintain the signal mutation phenomenon, there are some difficulties. Left prediction and right prediction, and crossover effective (CV) smoothing are studied, and two constraints in CV smoothing are proved. This paper proposes a finite prediction mixed-CV smoothing method to denoise, which can achieve the above objectives at the same time. The root signal characteristics of this method are analyzed. The method was tested by simulation and compared with finite mixed median filter (FMH) method and adaptive wavelet shrinkage method (WSA). The experimental results show that this method is feasible and better than FMH and WSA.