脑电_脑电信号分析_脑电分类_EEG_MATLABEEG_情绪分类_

%% 娓呯┖鐜鍙橀噺 clear; clc; close all; %% 娣诲姞褰撳墠璺緞鑷冲伐浣滅幆澧? currentFolder = pwd; addpath(genpath(currentFolder)); %% 鎻愬彇鍑轰竴缁勬牱鏈殑16涓俊閬撶殑鍘熷EEG淇″彿锛岃繘琛屾樉绀? %% 瀵煎叆涓ょ被鏍锋湰鏁版嵁浠ュ強瀵瑰簲鏍囩锛堜粠鍘熷鏁版嵁闆嗕腑鎻愬嚭锛? % 瀵瑰簲鏍囩璇存槑锛?浠ｈ〃awake鐘舵? 1浠ｈ〃sleep鐘舵? load('C:\Users\Administrator\Desktop\calm.mat') %鏁版嵁鐨勮矾寰勫彲淇敼 load('C:\Users\Administrator\Desktop\calm_label.mat') %鏁版嵁鐨勮矾寰勫彲淇敼 load('C:\Users\Administrator\Desktop\angry.mat') %鏁版嵁鐨勮矾寰勫彲淇敼 load('C:\Users\Administrator\Desktop\angry_label.mat') %鏁版嵁鐨勮矾寰勫彲淇敼 load('C:\Users\Administrator\Desktop\sad.mat') load('C:\Users\Administrator\Desktop\sad_label.mat') %% 灏忔尝鍙樻崲 calm_1 = zeros(900,31); %鍒嗛厤鍐呭瓨绌洪棿 angry_1 = zeros(800,31); sad_1 = zeros(1000,31); % 鍒嗗埆瀵逛袱绫绘暟鎹繘琛屽皬娉㈠彉鎹? for i=1:1:900 t = calm(i,:); [c,l] = wavedec(t,3,'db2'); calm_1(i,:) = c; end for j=1:1:800 q = angry(j,:); [c,l] = wavedec(q,3,'db2'); angry_1(j,:) = c; end for k=1:1:1000 p = sad(k,:); [c,l] = wavedec(p,3,'db2'); sad_1(k,:) = c; end % 灏嗚繘杩囧皬娉㈠彉鎹㈢殑涓ょ被鏁版嵁鍒嗗埆閫佸洖鍘熷彉閲? calm =calm_1; angry = angry_1; sad = sad_1; %% 灏嗘墍鏈夋暟鎹繘琛屽綊涓?寲澶勭悊 input_train = [calm;angry;sad]; [inputn,inputps] = mapminmax(input_train); % 褰掍竴鍖栧悗鐨勬暟鎹?鍥炲師鍙橀噺 calm = inputn((1:900),:); angry = inputn((901:1700),:); sad = inputn((1701:2700),:); %% 涓ょ被淇″彿鐨?0%鐨勬牱鏈敤浜庤缁僑VM锛?0%鐨勬牱鏈敤浜庢祴璇曟ā鍨嬪垎绫昏兘鍔? %浜х敓闅忔満鏁帮紝纭繚瀹為獙鐨勬棤搴忔? number_1 = randperm(size(calm,1)); number_2 = randperm(size(angry,1)); number_3 = randperm(size(sad,1)); % 鍖哄垎awake鐘舵?涓嬬殑鏍锋湰鏁版嵁 % 230*0.7=161涓猘wake鐘舵?鏍锋湰鐢ㄤ簬璁粌 % 230*0.3=69涓猘wake鐘舵?鏍锋湰鐢ㄤ簬娴嬭瘯 calm_train_label = zeros(size(calm_label,1)*0.7,size(calm_label,2)); calm_test_label = zeros(100,size(calm_label,2)); calm_train = zeros(size(calm,1)*0.7,size(calm,2)); calm_test = zeros(100,size(calm,2)); % 鍒嗛厤70%鐨刟wake鏁版嵁鑷宠缁冮泦 for i=1:1:(size(calm,1)*0.7) calm_train(i,:) = calm(number_1(i),:); calm_train_label(i,1) = calm_label(number_1(i),:); end % 鍒嗛厤30%鐨刟wake鏁版嵁(鍓╀綑鐨?鑷虫祴璇曢泦 for i=(size(calm,1)*0.7+1):1:size(calm,1)*0.7+100 calm_test(i-(size(calm,1)*0.7),:) = calm(number_1(i),:); calm_test_label(i-(size(calm,1)*0.7),1) = calm_label(number_1(i),:); end % 鍖哄垎鐫＄湢鐘舵?涓嬬殑鏍锋湰鏁版嵁 % 150*0.7=105涓猻leep鐘舵?鏍锋湰鐢ㄤ簬璁粌 % 150*0.3=45涓猻leep鐘舵?鏍锋湰鐢ㄤ簬娴嬭瘯 angry_train_label = zeros(size(angry_label,1)*0.7,size(angry_label,2)); angry_test_label = zeros(100,size(angry_label,2)); angry_train = zeros(size(angry,1)*0.7,size(angry,2)); angry_test = zeros(100,size(angry,2)); % 鍒嗛厤70%鐨剆leep鏁版嵁鑷宠缁冮泦 for j=1:1:size(angry,1)*0.7 angry_train(j,:) = angry(number_2(j),:); angry_train_label(j,1) = angry_label(number_2(j),:); end % 鍒嗛厤30%鐨剆leep鏁版嵁(鍓╀綑鐨?鑷虫祴璇曢泦 for j=(size(angry,1)*0.7+1):1:size(angry,1)*0.7+100 angry_test(j-(size(angry,1)*0.7),:)= angry(number_2(j),:); angry_test_label(j-(size(angry,1)*0.7),1) = angry_label(number_2(j),:); end sad_train_label = zeros(size(sad_label,1)*0.7,size(sad_label,2)); sad_test_label = zeros(100,size(sad_label,2)); sad_train = zeros(size(sad,1)*0.7,size(sad,2)); sad_test = zeros(100,size(sad,2)); % 鍒嗛厤70%鐨剆ad鏁版嵁鑷宠缁冮泦 for k=1:1:size(sad,1)*0.7 sad_train(k,:) = sad(number_3(k),:); sad_train_label(k,1) = sad_label(number_3(k),:); end % 鍒嗛厤30%鐨剆leep鏁版嵁(鍓╀綑鐨?鑷虫祴璇曢泦 for k=(size(sad,1)*0.7+1):1:size(sad,1)*0.7+100 sad_test(k-(size(sad,1)*0.7),:)= sad(number_3(k),:); sad_test_label(k-(size(sad,1)*0.7),1) = sad_label(number_3(k),:); end %% 姹囨?鏁版嵁 % 璇存槑锛歺Tr銆亂Tr瀵瑰簲涓哄叏浣撹缁冩牱鏈殑鏁版嵁銆佹爣绛? % 璇存槑锛歺Te銆亂Te瀵瑰簲涓哄叏浣撴祴璇曟牱鏈殑鏁版嵁銆佹爣绛? xTr = [angry_train;calm_train]; yTr = [angry_train_label;calm_train_label]; xTr1 = [sad_train;calm_train]; yTr1 = [sad_train_label;calm_train_label]; model = fitcsvm(xTr,yTr); % 璁粌xTr鏁版嵁锛屽嵆鐫＄潃浜嗗拰閱掔潃鐨勬暟鎹? model1 = fitcsvm(xTr1,yTr1); %%璁粌xTr1鏁版嵁锛屽嵆鎮蹭激鍜岄啋鐫?殑鏁版嵁 test = calm_test; yPre1 = predict(model,test); yPre3 = predict(model1,test); a = sum(calm_test_label == yPre1)/length(yPre1); b = sum(angry_test_label == yPre1)/length(yPre1); c = sum(sad_test_label == yPre3)/length(yPre3); disp(a); disp(b); disp(c); if(a<b) if(b<c) disp('the emotion is sad'); motion=c; else disp('the emotion is angry'); motion=b; end else if(a<c) disp('the emotion is sad'); motion=c; else disp('the emotion is calm'); motion=a; end end xTe = [angry_test;calm_test]; yTe = [angry_test_label;calm_test_label]; yPre = predict(model,xTe);% 灏嗘祴璇曟暟鎹?鍏ユā鍨嬶紝寰楀埌棰勬祴鏍囩yPre switch(motion) case a [y,Fs]=audioread('C:\Users\Administrator\Desktop\calm.mp3'); sound(y,Fs); case b [y,Fs]=audioread('C:\Users\Administrator\Desktop\angry.mp3'); sound(y,Fs); case c [y,Fs]=audioread('C:\Users\Administrator\Desktop\sad.mp3'); sound(y,Fs); end % %% 璁粌銆佸垎绫? % %% 瀹為獙缁撴灉鍒嗘瀽 % Acc = sum(yTe == yPre)/length(yTe); % C = confusionmat(yTe, yPre); % C涓烘贩娣嗙煩闃? % recall = 0; % accuracy = 0; % n = length(C); % for i=1:n % recall = recall + C(i,i)/sum(C(i,:)); % accuracy = accuracy + C(i,i)/sum(C(:,i)); % end % recall = recall/n; % accuracy = accuracy/n; % f1 = 2*(recall*accuracy)/(recall+accuracy); % disp(['Acc =' num2str(Acc)]); % disp(['f1 =' num2str(f1)]); % disp(['recall =' num2str(recall)]); % disp(['accuracy =' num2str(accuracy)]); % plotconfusion(yPre',yTe') % disp('222')