-
NS: noisy signal
-
S: original siganl
-
mean filter: ws = window size
-
median filter:
-
average filter: ns = number of noisy signal(different)
-
bandpass filter: l = low cut-off frequency, h = high ...
-
threshold filter: r = ratio(max abs(fft) / min ...)
-
wavelet filter: a = threshold
-
std filter:
-
NN: neural network
- why not pair them two or three together ?!
- (double noise) 1.mean 2.median 3.std -> unsatisfactory ![4](https://github.com/ZakiZhang168/8-methods-to-denoise/assets/130261283/fcd247ac-2ac5-4b5a-a08a-8bbc12904a4b)
- 4.Average -> better but unpractical ![5](https://github.com/ZakiZhang168/8-methods-to-denoise/assets/130261283/e10217e5-8e5f-47c2-b0a5-f43bed22307c)
- 5.Bandpass -> not bad, more smooth ![6](https://github.com/ZakiZhang168/8-methods-to-denoise/assets/130261283/82f973e5-283d-4109-b520-1b8417c77b7d)
- 6.Wavelet -> somewhat better than 5 ![7](https://github.com/ZakiZhang168/8-methods-to-denoise/assets/130261283/64dcfc30-c5e4-4e4f-b2ff-647d31cda029)
- 7.Threshold -> good ![8](https://github.com/ZakiZhang168/8-methods-to-denoise/assets/130261283/fec776c9-e47a-4874-8ff8-79a99ba6ef42)
- 8. NN -> not good, Difficult to evaluate ![9](https://github.com/ZakiZhang168/8-methods-to-denoise/assets/130261283/9a2651da-ffdd-4c85-8121-d4c5bc8c31af)
- Combination -> give up md,mn,std, but nn is useful ! ![10](https://github.com/ZakiZhang168/8-methods-to-denoise/assets/130261283/65c11d36-0dcd-4e01-9530-06b761616787)