MATLAB: Error while using foreground detection in matlab with webcam video

computer visionComputer Vision Toolboxforeground detectionimage processingImage Processing Toolbox

clear all
% video object to grab frame from live video
vidobj = imaq.VideoDevice('winvideo', 1);
detector = vision.ForegroundDetector(...
       'NumTrainingFrames', 50, ... % 
       'InitialVariance', 30*30); % initial standard deviation of 30
blob = vision.BlobAnalysis(...
       'CentroidOutputPort', false, 'AreaOutputPort', false, ...
       'BoundingBoxOutputPort', true, ...
       'MinimumBlobAreaSource', 'Property', 'MinimumBlobArea', 250);
   shapeInserter = vision.ShapeInserter('BorderColor','White');
   videoPlayer = vision.VideoPlayer();
for n=1:500
     f  = step(vidobj);
     frame = rgb2gray(f);
     fgMask = step(detector, frame);
     bbox   = step(blob, fgMask);
       %out    = step(shapeInserter, frame, fgMask); % draw bounding boxes 
       step(videoPlayer, fgMask); % view results in the video player
  end
  release(vidobj);
  release(videoPlayer);
This is the initial code I've written, I'm trying to display the masked frame, so it should come in black and white where the white ones are the foreground object. But all I get is a completely black output stream. Where have I gone wrong?

Best Answer

Hi Aravind,
This is probably because the 'InitialVariance' of the foreground detector is not appropriate for the frame datatype. The value of 30^2 works for 'uint8'. However, the default data type of the frame returned by the step() method of imaq.VideoDevice is 'single'. So you should either do
vidobj = imaq.VideoDevice('winvideo', 1, 'ReturnedDataType', 'uint8');
or set 'InitialVariance' to (30 / 255)^2.
Related Question