This paper discusses a technique for performing wavelet-based pattern
recognition. Known as 'progressive pattern recognition', this approach
is based on performing a subband decomposition of a target image usin
g the two-dimensional discrete wavelet transform and then performing p
artial correlations with different subband resolutions. This approach
offers the possibility of recognizing the image without having to perf
orm a full correlation of the image or of its complete wavelet transfo
rm, with the potential for considerable savings in computational time.
The algorithm for the progressive pattern recognition is presented wi
th results obtained from computer simulations of hand-print recognitio
n.