digital processing for eels data xiang yang watlabs, univeristy of waterloo

25
Digital Processing for Digital Processing for EELS Data EELS Data Xiang Yang Xiang Yang WATLABS, Univeristy of WATLABS, Univeristy of Waterloo Waterloo

Upload: loraine-riley

Post on 21-Jan-2016

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Digital Processing for EELS DataDigital Processing for EELS Data

Xiang YangXiang Yang

WATLABS, Univeristy of WaterlooWATLABS, Univeristy of Waterloo

Page 2: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Signals and NoiseSignals and Noise --1 --1

Signal: any Signal: any usefuluseful

informationinformation

Noise: any Noise: any unwatedunwated

informationinformation

Page 3: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Signals and NoiseSignals and Noise --2 --2

SSignalignal:: what you are measuring that is the result of what you are measuring that is the result of

the presence of your analytethe presence of your analyte

NoiseNoise: : extraneous information that can interfere with extraneous information that can interfere with

or alter the signal. or alter the signal.

Page 4: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Types of NoiseTypes of Noise --1 --1

Random Noise: sign & magnitude --unpredictableRandom Noise: sign & magnitude --unpredictable

Non-RandomNon-Random Noise Noise: :

sign & magnitude – correlated with some eventsign & magnitude – correlated with some event

Page 5: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Types of NoiseTypes of Noise --2 --2

Fundamental Noise: Fundamental Noise: ------- Due to the nature of light and matter------- Due to the nature of light and matter ------- Cannot be totally eliminated------- Cannot be totally eliminated

Non-FundamentalNon-Fundamental Noise Noise: : ------- Mostly due to instrumentation------- Mostly due to instrumentation ------- can be eliminated (theoretically)------- can be eliminated (theoretically)

Page 6: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

sX

deviation standardmean

NoiseSignal

Signal to Noise Ratio

(SNR)

Page 7: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo
Page 8: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Noise SourcesNoise Sources

Signal Source

Detector

Analog Treatments

Analog to DigitalConversion

Non-monochromate light source

Detector’s Dark Current, electromagnetic interference, etc.

Circuit noise, baseline, electromagnetic interference, etc.

Quantization effects

Page 9: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

SNR EnhancementSNR Enhancement HardwareHardware

Page 10: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Dwell Time v.s. SNRDwell Time v.s. SNR

Communication Communication between Computer & between Computer & MachineMachine

Page 11: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Ensemble AveragingEnsemble Averaging Collect multiple signals over Collect multiple signals over

the same time or wavelength the same time or wavelength ((x-axisx-axis) domain) domain

Calculate the mean signal at Calculate the mean signal at each point in the domaineach point in the domain

Re-plot the averaged signalRe-plot the averaged signal

Since noise is random (some Since noise is random (some +/ some -), this helps reduce +/ some -), this helps reduce the overall noise by the overall noise by cancellation!cancellation!

Page 12: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo
Page 13: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Boxcar AveragingBoxcar Averaging

– Take an average of Take an average of 2 or more signals in 2 or more signals in some domainsome domain

– Plot these points as Plot these points as the average signal in the average signal in the same domainthe same domain

– Can be done with Can be done with just one set of datajust one set of data

– You lose some detail You lose some detail in the overall signalin the overall signal

Page 14: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo
Page 15: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo
Page 16: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Digital FilteringDigital Filtering

Weighted Digital FilteringWeighted Digital Filtering

Fast Fourier Transform Digital FilteringFast Fourier Transform Digital Filtering

Page 17: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Weighted FilteringWeighted Filtering

Page 18: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo
Page 19: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Fast Fourier Transformation FilteringFast Fourier Transformation Filtering

Main Point: Noise is of a higher

frequency than the information

Page 20: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

FFT FilteringFFT Filtering

Noisy Data(Time Domain)

Tranformed Data(Frequency Domain)

FT

Modified Data

(Freq. Domain)

Low Pass Filter

Filtered Signal

FT

Page 21: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

FTFT

Filtering

Page 22: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

FFT ---- Real SampleFFT ---- Real Sample

Page 23: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

First Fourier Tranform

Cut off Frequency (0.003 Hz)

Page 24: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo
Page 25: Digital Processing for EELS Data Xiang Yang WATLABS, Univeristy of Waterloo

Thank You !