Unfortunately, in at least some versions, textscan() keeps reading for numeric fields as long as what is in the input is characters that could be part of a number format. It does not pay attention to the delimiter until after it has done that.
So the 48.2.81.56 is read as "4 is a digit, 8 is a digit, decimal place is part of a floating point format, 2 is a digit, decimal point has already occurred in the number so the number must have ended, 48.2; now look, the decimal point is the delimiter. Okay 8 and 1 are digits, decimal point is part of a floating point format, 5 and 6 are digits, end of string so the number has ended, 81.56. Now convert the 48.2 and 81.56 into the %u8 format."
What you should do instead is
integers = uint8( str2double( regexp( str, '\.', 'split' ) ) )
