Fwvq Compresses an image by vector quantizing its wavelet transform
Fwvq [-R Rate] [-m] [-s] [-u] [-r NLevel] [-b1 BiFilt1] [-b2 BiFilt2] [-o OrthoFilt] [-e EdgeFilt] image codebook
This macro compresses a 8 bits graylevel image using a vector quantization algorithm applied to the orthogonal/biorthogonal wavelet coefficients. It is based on the fwvq module, but is simpler to use.
One only has to specify the prefix codebook of all codebook sets and the input image to be compressed. The macro then looks for files named codebook.cb, codebook_x.cb, codebook_y.cb, codebook_q.cb, codebook_xq.cb, codebook_yq.cb, codebook_qr.cb and codebook_xqr.cb, and, if it finds them, uses them as input for fwvq. These files correspond respectively to CodeBook1, CodeBook2, CodeBook3, ResCodeBook1, ResCodeBook2, ResCodeBook3, ResResCodeBook1 and ResResCodeBook2. However, if one uses the Fwlbg_adap macro, one does not have to bother about these files. They are all automatically generated by this macro. It suffices to put the same prefix name codebook as argument of Fwlbg_adap and Fwvq (see documentation of Fwlbg_adap). Note that Fwvq need at least the file codebook.cb (CodeBook1) in order to work.
If the -R option is selected, then both a compressed file and a quantized image (which can be reconstructed from the compressed file) are generated. The compression is made in order to reach the target bit rate Rate. If the input image file has the form *.rim or *.img, then the compressed file is named *_Ratec.comp and the quantized image file is named *_Rateq.rim.
Details about wavelet transform are the same as for Fwlbg_adap macro. The reader is refered to its documentation for information about the -b1, -b2, -o and -e options.
Notice that the dimensions of the image should have a minimum of factor 2 in their decomposition in prime numbers. If not then a part of the image as large as possible is extracted and having the required number of factor 2.
The -r, -s, -u and -R options work as for fwvq module. If the -R option is not selected, then the rate distortion curve is computed (as if the -d option was selected in fwvq), see [GG92].
The -m option works slightly differently. If not selected, then it performs an approximative memory allocation (equivalent to -m 1 in fwvq). If selected, then it performs an exact memory allocation (equivalent to -m 2 in fwvq).
No scaling of wavelet coefficients is made. In other words,
the -w option is not activated when fwvq is called.
Last Modification date : Feb 11 2002