home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.dsp:1827 sci.math:9546 sci.math.num-analysis:2291
- Newsgroups: comp.dsp,sci.math,sci.math.num-analysis
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!wupost!gumby!destroyer!ubc-cs!unixg.ubc.ca!physics.ubc.ca!gill
- From: gill@physics.ubc.ca (Arnold Gill)
- Subject: Re: Help: Battle-Lemarie Wavelets
- Message-ID: <gill.712090374@physics.ubc.ca>
- Sender: news@unixg.ubc.ca (Usenet News Maintenance)
- Nntp-Posting-Host: physics.ubc.ca
- Organization: University of British Columbia, Vancouver, B.C., Canada
- References: <1992Jul24.223324.2869@vu-vlsi.vill.edu>
- Date: Sat, 25 Jul 1992 18:52:54 GMT
- Lines: 19
-
- pappu@vu-vlsi.vill.edu (Strange Attractor) writes:
-
- >I have been trying to generate the Battle-Lemarie wavelet and scaling functions
- >I have had some problems in implementing the algorithm given in Mallat's
- >paper in that I do not get enough points in the basic scaling function
- >phi(0,0). When I dilate this function, I am not able to evaluate the dilated
- >function wherever I like because of the lack of points.
-
- I think this may be a generic problem whenever you use a discrete wavelet
- rather than a continuous one. Eventually, your wavelet becomes too big
- for the data and there just aren't enough points left. If you have a
- continuous function describing the wavelet, you can always add in the
- missing pieces.
-
- By the way, don't you just use the same number of coefficients, but
- sample the data stream at larger intervals when dilating the wavelet?
-
- --
- Arnold Gill --- astrophysician trainee in exile gill@physics.ubc.ca
-