The red component of a color is stored as a number between 0-255, but "packed" into an integer. Here's my ASCII art, in binary, 32 bits per integer, showing the 8 bits each for Alpha,Red,Green,and Blue. Read the bit number vertically! I'm numbering from 1, not 0. (Also, you can argue that the numbers go the wrong direction - they are there just to help you count!)
11111111112222222222333
12345678901234567890123456789012
AAAAAAAARRRRRRRRGGGGGGGGBBBBBBBB the pixel value
The pixel >> 16
will shift the result right by 16 bits. In effect, this removes the G and B stuff
AAAAAAAARRRRRRRRGGGGGGGGBBBBBBBB move everything 16 to the right, this becomes
----------------AAAAAAAARRRRRRRR
where the - depends on sign. You can look it up but in this case it doesn't matter.
So, you have your Red number, almost. There is still the Alpha component, all that AA.... If you mask ("and") with 0xFF, which is the lowest 8 bits
----------------AAAAAAAARRRRRRRR (result of the bit shift)
00000000000000000000000011111111 (0xFF)
you get
000000000000000000000000RRRRRRRR
which is the result desired, a number between 0 and 255.