Add files via upload

A_Main.m

@@ -0,0 +1,59 @@
+%  Electroencephalogram (EEG) Feature Extraction toolbox
+
+%---Input-------------------------------------------------------------
+% X    : EEG signal (1 x samples)
+% opts : parameter settings
+% 
+
+%---Output------------------------------------------------------------
+% feat:  Feature vector 
+%---------------------------------------------------------------------
+
+
+%% Generate a sample random signal X
+fs = 500;             % Sampling frequency 
+Ts = 1 / fs;          % Period
+t  = 0 : Ts : 0.5; 
+X  = 0.01 * (sin(2 * pi * fs * t) + randn(1, length(t)));
+
+% Plot sample signal
+plot(t,X);  grid on
+xlabel('Number of samples');
+ylabel('Amplitude');
+
+
+%% Example 1 : Extract 3 normal features ( without parameter ) 
+% Generate a sample random signal X
+fs = 500;             % Sampling frequency 
+Ts = 1 / fs;          % Period
+t  = 0 : Ts : 0.5; 
+X  = 0.01 * (sin(2 * pi * fs * t) + randn(1, length(t)));
+
+% Hjorth Activity
+f1 = jfeeg('ha', X); 
+% Hjorth Mobility
+f2 = jfeeg('hm', X); 
+% Hjorth Complexity
+f3 = jfeeg('hc', X); 
+
+% Feature vector
+feat = [f1, f2, f3];
+
+
+%% Example 2 : Extract 2 features with parameter   
+% Generate a sample random signal X
+fs = 500;             % Sampling frequency 
+Ts = 1 / fs;          % Period
+t  = 0 : Ts : 0.5; 
+X  = 0.01 * (sin(2 * pi * fs * t) + randn(1, length(t)));
+
+% Band Power Alpha
+opts.fs = 500;
+f1 = jfeeg('bpa', X, opts); 
+% Tsallis Entropy
+opts.alpha = 2;
+f2 = jfeeg('te', X, opts);
+
+% Feature vector
+feat = [f1, f2];
+

jArithmeticMean.m

@@ -0,0 +1,7 @@
+%[2014]-"A Comparative Study on Classification of Sleep Stage Based on
+%EEG Signals Using Feature Selection and Classification Algorithms" (4)
+
+function AM = jArithmeticMean(X,~)
+AM = mean(X);
+end
+

jAutoRegressiveModel.m

@@ -0,0 +1,14 @@
+%[2012]-"Classification of EMG signals using combined features and 
+%soft computing techniques" (1-2)
+
+function AR = jAutoRegressiveModel(X,opts)
+% Parameter
+order = 4;    % order
+
+if isfield(opts,'order'), order = opts.order; end
+
+Y  = arburg(X,order); 
+% First index is meaningless
+AR = Y(2 : order + 1); 
+end
+

jBandPowerAlpha.m

@@ -0,0 +1,13 @@
+
+function BPA = jBandPowerAlpha(X,opts)
+% Parameters         
+f_low  = 8;      % 8 Hz
+f_high = 12;     % 12 Hz
+
+% sampling frequency 
+if isfield(opts,'fs'), fs = opts.fs; end
+
+% Band power 
+BPA = bandpower(X, fs, [f_low f_high]);
+end
+

jBandPowerBeta.m

@@ -0,0 +1,13 @@
+
+function BPB = jBandPowerBeta(X,opts)
+% Parameters         
+f_low  = 12;      % 12 Hz
+f_high = 30;      % 30 Hz
+
+% sampling frequency 
+if isfield(opts,'fs'), fs = opts.fs; end
+
+% Band power 
+BPB = bandpower(X, fs, [f_low f_high]);
+end
+

jBandPowerDelta.m

@@ -0,0 +1,13 @@
+
+function BPD = jBandPowerDelta(X,opts)
+% Parameters         
+f_low  = 1;      % 1 Hz
+f_high = 4;      % 4 Hz
+
+% sampling frequency 
+if isfield(opts,'fs'), fs = opts.fs; end
+
+% Band power 
+BPD = bandpower(X, fs, [f_low f_high]);
+end
+

jBandPowerGamma.m

@@ -0,0 +1,13 @@
+
+function BPG = jBandPowerGamma(X,opts)
+% Parameters         
+f_low  = 30;      % 30 Hz
+f_high = 64;      % 64 Hz
+
+% sampling frequency 
+if isfield(opts,'fs'), fs = opts.fs; end
+
+% Band power 
+BPG = bandpower(X, fs, [f_low f_high]);
+end
+

jBandPowerTheta.m

@@ -0,0 +1,13 @@
+
+function BPT = jBandPowerTheta(X,opts)
+% Parameters         
+f_low  = 4;      % 4 Hz
+f_high = 8;      % 8 Hz
+
+% sampling frequency 
+if isfield(opts,'fs'), fs = opts.fs; end
+
+% Band power 
+BPT = bandpower(X, fs, [f_low f_high]);
+end
+

jFirstDifference.m

@@ -0,0 +1,11 @@
+%[2014]-"Feature Extraction and Selection for Emotion Recognition from
+%EEG"
+
+function FD = jFirstDifference(X,~)
+T = length(X);
+Y = 0;
+for t = 1 : T - 1
+  Y = Y + abs(X(t+1) - X(t));
+end
+FD = (1 / (T - 1)) * Y;
+end

jHjorthActivity.m

@@ -0,0 +1,7 @@
+%[2018]-"A Novel Multi-Class EEG-Based Sleep Stage Classification System"
+%(3)
+
+function HA = jHjorthActivity(X,~) 
+sd = std(X); 
+HA = sd ^ 2;
+end

jHjorthComplexity.m

@@ -0,0 +1,16 @@
+%[2018]-"A Novel Multi-Class EEG-Based Sleep Stage Classification System"
+%(5)
+
+function HC = jHjorthComplexity(X,~)
+% First & second derivative
+x0  = X(:);
+x1  = diff([0; x0]);
+x2  = diff([0; x1]);
+% Standard deviation of first & second derivative 
+sd0 = std(x0);
+sd1 = std(x1);
+sd2 = std(x2); 
+% Complexity
+HC  = (sd2 / sd1) / (sd1 / sd0);
+end
+

jHjorthMobility.m

@@ -0,0 +1,13 @@
+%[2018]-"A Novel Multi-Class EEG-Based Sleep Stage Classification System"
+%(4)
+
+function HM = jHjorthMobility(X,~)
+% First derivative
+x0  = X(:);
+x1  = diff([0; x0]); 
+% Standard deviation 
+sd0 = std(x0); 
+sd1 = std(x1);
+HM  = sd1 / sd0;
+end
+

jKurtosis.m

@@ -0,0 +1,5 @@
+
+function KURT = jKurtosis(X,~)
+% Kurtosis 
+KURT = kurtosis(X);
+end

jLogEnergyEntropy.m

@@ -0,0 +1,8 @@
+%[2017]-"Automated detection of focal EEG signals using features 
+%extracted from flexible analytic wavelet transform" (8)
+
+function LogEn = jLogEnergyEntropy(X,~)
+% Entropy 
+LogEn = sum(log(X .^ 2)); 
+end
+

jLogRootSumOfSequentialVariation.m

@@ -0,0 +1,11 @@
+%[2018]-"A Novel Multi-Class EEG-Based Sleep Stage Classification 
+%System" (11)
+
+function LRSSV = jLogRootSumOfSequentialVariation(X,~)
+N = length(X); 
+Y = zeros(1, N-1);
+for i = 2:N
+  Y(i-1) = (X(i) - X(i-1)) ^ 2;
+end
+LRSSV = log10(sqrt(sum(Y)));
+end

jMaximum.m

@@ -0,0 +1,7 @@
+%[2014]-"A Comparative Study on Classification of Sleep Stage Based on EEG
+%Signals Using Feature Selection and Classification Algorithms" (2)
+
+function X_max = jMaximum(X,~)
+X_max = max(X);
+end
+

jMeanCurveLength.m

@@ -0,0 +1,12 @@
+%[2014]-"A Comparative Study on Classification of Sleep Stage Based on
+%EEG Signals Using Feature Selection and Classification Algorithms" (13)
+
+function MCL = jMeanCurveLength(X,~)
+N = length(X); 
+Y = 0;
+for m = 2:N
+  Y = Y + abs(X(m) - X(m-1));
+end
+MCL = (1 / N) * Y;
+end
+

jMeanEnergy.m

@@ -0,0 +1,6 @@
+%[2014]-"A Comparative Study on Classification of Sleep Stage Based on
+%EEG Signals Using Feature Selection and Classification Algorithms" (12)
+
+function ME = jMeanEnergy(X,~)
+ME = mean(X .^ 2);
+end

jMeanTeagerEnergy.m

@@ -0,0 +1,12 @@
+%[2014]-"A Comparative Study on Classification of Sleep Stage Based on
+%EEG Signals Using Feature Selection and Classification Algorithms" (11)
+
+function MTE = jMeanTeagerEnergy(X,~)
+N = length(X); 
+Y = 0;
+for m = 3:N
+  Y = Y + ((X(m-1) ^ 2) - X(m) * X(m-2));
+end
+MTE = (1 / N) * Y;
+end
+

jMedian.m

@@ -0,0 +1,7 @@
+%[2014]-"A Comparative Study on Classification of Sleep Stage Based on 
+%EEG Signals Using Feature Selection and Classification Algorithms" (8-9)
+
+function X_med = jMedian(X,~)
+X_med = median(X);
+end
+

jMinimum.m

@@ -0,0 +1,7 @@
+%[2014]-"A Comparative Study on Classification of Sleep Stage Based on EEG
+%Signals Using Feature Selection and Classification Algorithms" (1)
+
+function X_min = jMinimum(X,~)
+X_min = min(X);
+end
+

jNormalizedFirstDifference.m

@@ -0,0 +1,12 @@
+%[2014]-"Feature Extraction and Selection for Emotion Recognition from 
+%EEG"
+
+function NFD=jNormalizedFirstDifference(X,~)
+T = length(X); 
+Y = 0; 
+for t = 1 : T - 1
+  Y = Y + abs(X(t+1) - X(t));
+end
+FD  = (1 / (T - 1)) * Y; 
+NFD = FD / std(X);
+end

jNormalizedSecondDifference.m

@@ -0,0 +1,12 @@
+%[2014]-"Feature Extraction and Selection for Emotion Recognition from 
+%EEG"
+
+function NSD = jNormalizedSecondDifference(X,~)
+T = length(X); 
+Y = 0;
+for t = 1 : T - 2
+  Y = Y + abs(X(t+2) - X(t));
+end
+SD  = (1 / (T - 2)) * Y; 
+NSD = SD / std(X);
+end

jRatioBandPowerAlphaBeta.m

@@ -0,0 +1,20 @@
+
+function RBA = jRatioBandPowerAlphaBeta(X,opts)
+% Parameters         
+f_low1  = 8;       % 8 Hz
+f_high1 = 12;      % 12 Hz
+
+f_low2  = 12;      % 12 Hz
+f_high2 = 30;      % 30 Hz
+
+% sampling frequency 
+if isfield(opts,'fs'), fs = opts.fs; end
+
+% Band power alpha
+BPA = bandpower(X, fs, [f_low1 f_high1]);
+% Band power beta
+BPB = bandpower(X, fs, [f_low2 f_high2]);
+% Ratio beta/alpha
+RBA = BPB / BPA;
+end
+

jRenyiEntropy.m

@@ -0,0 +1,16 @@
+%[2018]-"Improved binary dragonfly optimization algorithm and wavelet
+%packet based non-linear features for infant cry classification" (4)
+
+function ReEn = jRenyiEntropy(X,opts) 
+% Parameter
+alpha = 2;     % alpha
+
+if isfield(opts,'alpha'), alpha = opts.alpha; end
+
+% Convert probability using energy
+P    = (X .^ 2) ./ sum(X .^ 2);
+% Entropy 
+En   = P .^ alpha; 
+ReEn = (1 / (1 - alpha)) * log2(sum(En)); 
+end
+

jSecondDifference.m

@@ -0,0 +1,11 @@
+%[2014]-"Feature Extraction and Selection for Emotion Recognition from
+%EEG"
+
+function SD = jSecondDifference(X,~)
+T = length(X); 
+Y = 0;
+for t = 1 : T - 2
+  Y = Y + abs(X(t+2) - X(t));
+end
+SD = (1 / (T - 2)) * Y;
+end

jShannonEntropy.m

@@ -0,0 +1,11 @@
+%[2018]-"Improved binary dragonfly optimization algorithm and wavelet
+%packet based non-linear features for infant cry classification" (3)
+
+function ShEn = jShannonEntropy(X,~) 
+% Convert probability using energy
+P    = (X .^ 2) ./ sum(X .^ 2);
+% Entropy 
+En   = P .* log2(P);
+ShEn = -sum(En); 
+end
+

jSkewness.m

@@ -0,0 +1,5 @@
+
+function SKEW = jSkewness(X,~)
+SKEW = skewness(X);
+end
+

jStandardDeviation.m

@@ -0,0 +1,7 @@
+
+function SD = jStandardDeviation(X,~)
+N  = length(X); 
+mu = mean(X); 
+SD = sqrt((1 / (N - 1)) * sum((X - mu) .^ 2));
+end
+