Abstract
A novel border-following algorithm is used to reconstruct the borders
and missing links of noisy and broken handwritten digits. When combined
with a suitable rejection threshold, improved performance from handwritten
character classifiers is achieved. Experimental results are given
which show the effectiveness of the method, raising recognition rates
by an additional 5% for handwritten digits in the presence of gaps
and noise.
Users
Please
log in to take part in the discussion (add own reviews or comments).