|
This thesis initially introduces the application of
multiresolutionanalysis of wavelet transform upon image signal,
further, achieves our goalby using components of different
resolutions, such as edge detection and imagecompression. In the
second session, we will impose extrapolate algorithm onthe
coefficient matrix of wavelet transform in order to cancel the
boundaryeffect caused by circular convolution in the
decomposition and reconstructionof digital image signal.
Nonstationary signals processing emerges to be an important
issue now andagain. For its window frame that full of
flexibility in the time-scaleplane, multiresolution analysis of
wavelet transform becomes a fine tool inthe signal processing.
Since the width of channel band is limited, thetopic on how to
achieve high image quality from minimum data becomingquite
important in the image processing. Because of the three
dataredundancies that frequently existed, they are coding
redundancy, interpixelredundancy and psychovisual redundancy,
wavelet transform capable totake off interpixel and psychovisual
redundancies easily with itsmultiresolution analysis. In
addition, since important data mostly existedin the edge of
image in general, the coding redundancy can be reducedwith
accordance to the modulus magnitude and angle under
differentscales. Furthermore, the compression ratio is also
upgraded.
|