the baseline like in [1] or [9] was not adopted, because the second feature vector is mainly necessary to distinguish between upper- and lowercase letters and descenders, and the simple baseline derivation was sufficiently precise for that purpose.
the baseline like in [1] or [9] was not adopted, because the second feature vector is mainly necessary to distinguish between upper- and lowercase letters and descenders, and the simple baseline derivation was sufficiently precise for that purpose.