If you put this line after your line plotting, you'll see the basics of using the patch command:
patch([x1; x1(1)],[y1; y1(1)],'g')
The first two arguments are the (x,y) coordinates of the polygon that you want to fill. Notice how I "closed the loop" by appending the initial x1 and y1 points again, at the end of the vector. Here is the result:
I realize that that is not the exact area you want filled. But it gives you the idea.
To do what you really want, you'll need to similarly trace the outline of each of the polygons you want filled in, based on your input vectors. You'll need to combine some pieces of both x1 and x2 as input to patch, for some regions.