median

中值滤波（MedianFilter）

中值滤波（MedianFilter）

中值滤波的中⼼思想就是逐项地遍历信号，并⽤相邻信号项的中值替换当前值。这种⽅法是的滤波处理⾮常快速，⽽且对于⼀维数据集合和

⼆维数据集合（例如图像）都适⽤。

在下⾯的例⼦中，我们使⽤SciPy信号⼯具箱中的实现。

#!/usr/bin/python3.4

#_*_encode:utf-8_*_

importnumpyasnp

importpylabasp

assignal

#getsomelineardata

x=ce(0,3,101)

print('Withoutnoisy:',x)

#addsomenoisysignal

x[5::10]=1.5

print('Withnoisy:',x)

(x)

(t(x,3))

(t(x,5))

(['originalsignal','length3','length5'])

()

运⾏⽂本输出结果如下：

Withoutnoisy:[0.0.030.060.090.120.150.180.210.240.270.30.33

0.360.390.420.450.480.510.540.570.60.630.660.69

0.720.750.780.810.840.870.90.930.960.991.021.05

1.081.111.141.171.21.231.261.291.321.351.381.41

1.441.471.51.531.561.591.621.651.681.711.741.77

1.81.831.861.891.921.951.982.012.042.072.12.13

2.162.192.222.252.282.312.342.372.42.432.462.49

2.522.552.582.612.642.672.72.732.762.792.822.85

2.882.912.942.973.]

Withnoisy:[0.0.030.060.090.121.50.180.210.240.270.30.33

0.360.390.421.50.480.510.540.570.60.630.660.69

0.721.50.780.810.840.870.90.930.960.991.021.5

1.081.111.141.171.21.231.261.291.321.51.381.41

1.441.471.51.531.561.591.621.51.681.711.741.77

1.81.831.861.891.921.51.982.012.042.072.12.13

2.162.192.221.52.282.312.342.372.42.432.462.49

2.521.52.582.612.642.672.72.732.762.792.821.5

2.882.912.942.973.]

中值滤波，窗⼝长度为3和5：

从图形中可以看出，窗⼝越⼤，信号和原始信号相⽐失真越严重，但同时看上去也越平滑。