$1.$ For this lab you are going to implement a background subtraction solution for the included video

clip.

$2.$ Since we$$ve been mainly working with images up to this point, we$$ll keep things simple and

break down our video into several images. Recall that a video is really a sequence of images that

are rapidly displayed one after another. The following matlab code will break your video into

multiple image files:

mov $=$ VideoReader$(\text{'}$YourVideoFileName$.$wmv$\text{'})$;

for i$=1$:mov.NumberOfFrames

img $=$ read$($mov$,$i$)$;

bwImage $=$ rgb$2$gray$($img$)$;

outputFileName $=$ sprintf$(\text{'}$bwImage$\%$d$.$jpg$\text{'},$i$)$;

imwrite$($bwImage$,$ outputFileName, $\text{'}$jpg$\text{'})$;

end

$3.$ You will want to run the above code only once on your video file. This step is typically called

preprocessing.

$4.$ There are several different methods available for background subtraction, and we$$re going to

focus on a basic approach. A basic approach is to track the last $10$ pixel values of each video

frame. If the intensity of the pixel changes either brighter or darker by a set threshold, we will

flag this pixel as a foreground pixel. In order to do this, you will need to know the dimensions of

the video you will be using and create a matrix of those dimensions with a third dimension of

value $10.$ For example, if the video you are using is $1920$x$1080,$ you will want to create a matrix

of size $1920$x$1080$x$10.$ A matrix of this size will allow us to keep track of the last $10$ values for

each pixel.

$5.$ If you$$re not sure on the size of your video, you can read in one of the image files and check it$$s

dimensions:

myTest $=$ imread$(\text{'}$bwImage$1.$jpg$\text{'})$;

size$($myTest$)$

ans $=$

$10801920$

$6.$ The image above is $1920$ by $1080,$ but Matlab flips the horizontal and vertical so it displays it as

shown above.

$7.$ Read in the first $10$ frames to create a matrix that you will track the last $10$ frames viewed. This

matrix will be used to calculate the average value at each pixel location.

$8.$ What you want to do next is create a loop that will run through each of the image files that you

created. In each of these iterations you will want to do the following:

a$.$ Compare each of the pixel values $($height by width$)$ to the average value stored in the

average matrix. If the absolute value of the difference is greater than some threshold

$($start with threshold $=20)$ set the value of that correspond pixel in the newImage to be

white $($intensity value $255).$ If it$$s less than the threshold you will set the value of that

corresponding pixel in the newImage to be black $($intensity value $0).$ If the pixel value is

white, it is a foreground pixel, if it$$s set to black, it$$s a background pixel.

b$.$ Update the average matrix by replacing the oldest image in the matrix with the current

image.

$9.$ When you$$ve created a new set of images, you$$ll want to combine them back together again

$($Please note you must first create a directory called $$modifiedImages$).$

mov $=$ VideoReader$(\text{'}$baseball$.$mp$4\text{'})$;

totalImages $=360$;

videoFrameRate $=$ mov.FrameRate;

videoHeight $=$ mov.Height;

videoWidth $=$ mov.Width;

outputVideo $=$ VideoWriter$(\text{'}$myfile$.$avi'$)$;

outputVideo.FrameRate $=$ videoFrameRate;

open$($outputVideo$)$;

for i$=11$:totalImages

outputFileName $=$ sprintf$(\text{'}$modifiedImages$\backslash \backslash$bwImage$\%$d$.$jpg$\text{'},$ i$)$;

myImage $=$ imread$($outputFileName$)$;

writeVideo$($outputVideo$,$ myImage$)$;

end

close$($outputVideo$)$