where h( ) k is a expansion coefficient, and then a wavelet
dyadic discrete family ( ) 2 2 2
j
j ψ t k ⎧ ⎫ ⎨ − ⎬ ⎩ ⎭
can be formed as
an orthonornal basis with consisting of finite-energy signals.
Thus, the gait data, assumed as an arbitrary function
( ) ( ) 2 S t ∈ L R of length , can be decomposed by using the
above wavelet family formed ,and the decomposition is able
to be inverted. In the process of the decomposition, the gait
data can be decomposed into two subsingnals at each resolution
level, averaged and detail signal, described by using the scaling
and wavelet expansion as follows: