how to iterate through all the pixels within a bbox returned by vision.BlobAnalysis

Hi I want to iterate through all the pixels within a bbox returned by vision.BlobAnalysis to sum their values, tried following code assuming (x,y) are the coordinates for first pixel, width is number of columns moving right and height is number of rows moving down (seems like I was wrong get an error indicating negative values as k,j coordinates). Can someone kindly explain a bit.

sum=0;
  for k=track.bbox(1,1):track.bbox(1,1)+tracks.bbox(1,3)
       for j=tracks.bbox(1,2):tracks.bbox(1,2)+tracks.bbox(1,4)
               sum=sum+img(k,j);
       end
   end

Answer 1

[Walter Roberson]

   %remember, bounding box is _x_ first and _x_ is column not row!

   fr = track.bbox(1,2);
   fc = track.bbox(1,1);
   lr = fr + track.bbox(1,4) - 1;   %height includes first pixel
   lc = fc + track.bbox(1,3) - 1;   %width includes first pixel
   pixtotal = sum( sum( img(fr:lr, fc:lc) ) );

and never use "sum" as the name of a variable: it interferes with using sum() as the important MATLAB routine

Note: if you are adding the pixel values only in order to find the mean pixel value, then do not bother adding them:

pixmean = mean2( img(fr:lr, fc:lc) );

[Maria Kanwal]

Thanks Walter but I still get the error: "Subscript indices must either be real positive integers or logicals."

[Walter Roberson]

What does track.bbox show up as?

Please show your code for creating track.bbox as there are a number of different bounding box detectors.

[Maria Kanwal]

bbox shows up as a struct with 4 fields x,y,width and height.

I am using matlab's multiObjectTracking which is using vision.BlobAnalysis for getting bboxes after foreground detection. The problem is that some of the x and y coordinates have negative values so can not index pixels in the image which cause the error, can't understand what coordinate system is working here, I guess it may be axes so tried to use axes2pix but its not working. Also there is some strange behaviour the example 2 outputs +ve values for negative inputs but example 1 outputs -ve values for negative inputs.