In many applications, wavelets are usually expected to have the following p
roperties: compact support, orthogonality, linear-phase, regularity, and in
terpolation, To construct such wavelets, it is crucial designing scaling fu
nctions with the above properties. In two- and three-band cases, except for
the Haar functions, there exists no scaling function with the above five p
roperties, In M-band case (M greater than or equal to 4), more free degrees
available in design enable us to construct such scaling functions. In this
paper, a novel approach to designing such scaling functions is proposed. F
irst, we extend the two-band Dubuc filters to M-band case. Next, the M-band
FIR regular symmetric interpolating scaling filters are parameterized, and
then, M-band FIR regular orthogonal symmetric interpolating scaling filter
s (OSISFs) are designed via optimal selection of parameters. Finally, two f
amily of four-band and five-band OSISFs and scaling functions are developed
, and their smoothnesses are estimated.