Two approaches are proposed for Chinese signature verification in this arti
cle. The topological approach, which extracts the static topological featur
es of the Chinese signature characters such as blocks, holes, segments, end
-points, and cross-points, is first presented. Based on this approach, an a
utomatic Chinese signature verification system is designed and implemented.
Experimental results demonstrate a recognition rate of 94%. Two new algori
thms for this approach are proposed in this article. The waveform matching
approach to measure the similarity of waveforms is then presented. This app
roach cuts the waveforms into a string of valley and peak (V-P-V) units and
calculates the distance between two strings by using insertion, deletion,
and substitution to obtain the similarity of two waveforms. The time comple
xity of this algorithm is O(n x m), where n and m are the unit numbers of t
he two waveform strings, respectively. The V-P-V algorithm used in a Chines
e signature verification system achieves a recognition rate of 90%. (C) 199
9 John Wiley & Sons, Inc.