How to classify the human postures of sitting and standing from video input?
1 Ansicht (letzte 30 Tage)
Ältere Kommentare anzeigen
Currently i have done human detection on input from webcam. It puts a rectangle bounding box around the person and the centroid of the box is shown too. How to proceed further to detect the sitting and standing posture?
vid = videoinput('winvideo', 1);
for i=1:10
misc = getsnapshot(vid);
end
bg = misc;
X = im2double(misc);
misc = (X(:,:,1)+X(:,:,2)+X(:,:,3));
misc(misc==0) = 0.001;
X(:,:,1) = X(:,:,1)./misc;
X(:,:,2) = X(:,:,2)./misc;
DlgH = figure('units','normalized','outerposition',[0 0 1 1]);
H = uicontrol('Style', 'PushButton', ...
'String', 'Close', ...
'Callback', 'delete(gcbf)');
while (ishandle(H))
mov = getsnapshot(vid);
subplot(1,2,1), imshow(mov); %1. show current frame%
title('Current Frame');
X1 = im2double(mov);
misc = (X1(:,:,1)+X1(:,:,2)+X1(:,:,3));
misc(misc==0) = 0.001;
X1(:,:,1) = X1(:,:,1)./misc;
X1(:,:,2) = X1(:,:,2)./misc;
A = imabsdiff (X,X1);
A (A<0.27) = 0; %threshold for background subtraction figure 2%
A (A>=0.27) = 1;
Res = imdilate (A,ones(9));
GX = rgb2gray(bg);
GX1 = rgb2gray(mov);
FD = im2double(imabsdiff(GX,GX1));
%subplot(2,2,2), imshow(FD); %2. show foreground%
title('Foreground');
FD(FD<(0.10)) = 0; %threshold for figure 3%
FD(FD>=(0.10)) = 1;
DR = imdilate(FD,ones(9));
re = imfill(DR,'holes');
finalRes(:,:,1) = Res(:,:,1).*re;
finalRes(:,:,2) = Res(:,:,2).*re;
finalRes(:,:,3) = Res(:,:,3).*re;
%subplot(2,2,3), imshow(finalRes); %3. show processed foreground%
title('Foreground2');
boundaries = bwboundaries(rgb2gray(imclose(finalRes,ones(3))));
bndrSize = size(boundaries);
% subplot(2,2,1:2)
subplot(1,2,2), imshow(mov); %4. show final frame%
title('Human detection');
hold on
for k = 1:bndrSize(1)
bnd = boundaries{k};
RB2 = max(bnd);
LT2 = min(bnd);
w = RB2(2)-LT2(2);
h = RB2(1)-LT2(1);
x1 = LT2(2);
y1 = LT2(1);
x2 = LT2(2)+w;
y2 = LT2(1)+h;
x3 = (x1 +x2)/2;
y3 = (y1 + y2)/2;
if (max(RB2-LT2)>100) %with rectangle over them. adjust this to cover only large objects, small values will cover smaller objects %
rectangle('Position',[LT2(2),LT2(1),RB2(2)-LT2(2),RB2(1)-LT2(1)],'EdgeColor','g','LineWidth',1);
rectangle('Position',[x3,y3,10,10],'EdgeColor','r','LineWidth',1);
end
end
end
close
0 Kommentare
Antworten (0)
Siehe auch
Kategorien
Mehr zu Visualization finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!