The directories L1 and L2 contain the results of applying our wavelet
compression program to the images in ../pgmfiles.  The files in L1 were
obtained by setting the parameters to minimize the L1 (mean-absolute)
error, while the files in L2 were obtained by setting the parameters to
minimize the L2 (root-mean-square) error.


Each file has the following form (this is L2/img0001):

2003454 0.3931931 0.6380734 3.0000000
1270689 0.8070102 1.0934989 5.0000000
651647 1.3605550 1.7799067 11.0000000
330608 2.0511584 2.6848526 19.0000000
172087 3.0354803 3.9995740 35.0000000
82839 4.4191855 5.8731666 59.0000000
34487 6.1724211 8.3160758 88.0000000
13061 8.2520905 11.2189112 111.0000000
4459 10.5183352 14.3165386 157.0000000
1560 12.8942335 17.3853183 185.0000000
691 14.9298882 20.1276994 195.0000000
369 17.0374708 22.6746970 196.0000000
325 18.7515586 24.2552808 198.0000000

The first figure in each line is the number of bytes in the compressed
file.  The next figure is the mean-absolute (average) error in units of
grey-scales.  The next figure is the root-mean-square error in
grey-scales, and the last figure is the maximum error in grey-scales.

For example, at a size of 651,647 bytes, the mean-absolute error is
1.36 grey scales, the RMS error is 1.78 greyscales, and the maximum
error is 11 greyscales.  Each of the original files contain 6,291,456
8-bit pixels.