Your last sentence has the answer: your data type is uint8. Dividing a uint8 by a uint8 is going to give you another integer, so you won't get any decimal places.
Your options are to convert to either double or single and do the operation, or as you said you can store each entry as a fraction. Since they are uint8, every entry of the matrix will contain the numerator, and every denominator is 98 -- so you already have all of the information.
