Understanding wavelet implementation

I need to implement the simple integer 5/3 wavelet, but need a little help
with code.

My understanding till now is that a 5/3 wavelet basically boils down to
following operations,

For even pixels, perform a prediction using a predictor based on average of
left and right pixels. For odd pixels, add half of average of errors at left
and right pixels. And then performing same operations in vertical direction.

Is this correct?

All the wavelet implementations I have found implement some kind of
generalised wavelet calculation method which makes it hard (for me) to
figure out how these operations are performed.

I am looking for clues which can either help me understand the available
wavelet code, or get me started on writing my own code.

How can both the horizontal and vertical transforms be implemented in a
single pass, without 'looking ahead'? Is this even possible (or required)?
While decompression we will only get decompressed pixels in raster-scan
order, so how will looking ahead work then?

I only want to perform 1-level transform, but can multiple level transforms
be done in a single pass?

J


.

Relevant Pages

  • Re: Understanding wavelet implementation
    ... My understanding till now is that a 5/3 wavelet basically boils down to ... perform a prediction using a predictor based on average of ... left and right pixels. ... The above implementation is the "lifting" implementation of a wavelet. ...
    (comp.compression)
  • Re: Understanding wavelet implementation
    ... My understanding till now is that a 5/3 wavelet basically boils down to ... For even pixels, perform a prediction using a predictor based on average ... The above implementation is the "lifting" implementation of a wavelet. ...
    (comp.compression)
  • Re: Help with 2d Wavelets
    ... read fairly widely on 2D transforms but remain unclear on a few ... different filters are applied to the columns of those subbands ... produced by the row transform rather than to the original image. ... I believe that multi-dimensionalwavelet ...
    (sci.math.num-analysis)
  • Re: Fourier transform obsolete?
    ... I ignored the "ringing" in order to concentrate on ... with wavelet transforms which don't seem to suffer from the ...
    (sci.physics)
  • Re: discrete log-based FFT
    ... You might want to look at wavelet transforms. ... They do something different than Fourier transforms, but work by octaves. ...
    (sci.physics.research)