This archive contains implementation of the Histogram Binning/Packing method, described in T.Bernas, P.Robinson, B.Rajwa, R.Starosolski "Application of detector precision characteristics and histogram packing for compression of biological fluorescence micrographs".


This implementation is intended for research purposes only; it is provided "as is"; authors make no warranty of any kind, either express or implied, with respect to this software. 
(c) 2011 Roman Starosolski and Tytus Bernas


User has to input in the commandline image dark signal and noise coefficients, based on them implementation outputs mappings of levels in noisy (original) image to levels after Histogram Binning (HB) and to levels after Histogram Binning and Packing (HBP). 
Implementation may also perform actual image conversions for HB and HBP methods. If filenames are given (input filename has to be in the PGM format, type P5 - single channel binary BigEndian), then one of following three modes of operation is selected: 
a) -raw_to_HB         perform histogram binning (default) of noisy image		
b) -raw_to_HBP        perform histogram binning and packing of noisy image
c) -HBP_to_HB         restore HB image from histogram packed image


Commandline parameters, any order, *-obligatory:
 -p_max_levels n    maximum intensity, default 65535
 -p_multn n         * multiplicative noise coefficient (square of)
 -p_poissn n        * Poisson noise coefficient (square of)
 -p_addn n          * Additive noise coefficient (variance)
 -p_darklev n       * Dark signal (average intensity)
 -p_conf_lev n      confidence level (t value), default 1.96

 -f_in f            input file name, if not given just output HB levels
                    image must be in PGM P5 single channel binary BigEndian
 -f_out f           otput file name

 -raw_to_HB         perform histogram binning (default)
 -raw_to_HBP        perform histogram binning and packing
 -HBP_to_HB         restore histogram packed image


This archive contains files:
readme.txt 	- this file
hbp.c		- implementation in C
hbp.exe		- Win32 executable
mp_fib_001.pgm	- noisy microscope image, detector characteristic: p_multn=0, p_poissn=0.441128967, p_addn=44.10389456, p_darklev=17.83913285
mp_fib_001_hb.pgm   - image mp_fib_001.pgm after histogram binning with above parameters
mp_fib_001_hbp.pgm  - image mp_fib_001.pgm after histogram binning and packing with above parameters


Usage examples:


>hbp -p_max_levels 4095 -p_multn 0 -p_poissn 0.441128967 -p_addn 44.10389456 -p_darklev 17.83913285 

this will output mapping of levels betwen raw(noisy), histogram Binned (HB) and Histogram Binned and Packed (HBP) image:

p_max_levels   =4095.000000
p_multn        =0.000000
p_poissn       =0.441129
p_addn         =44.103895
p_darklev      =17.839133
p_conf_lev     =1.960000

processing mode  : raw->HB

   raw level range      HB level    HBP level

[      0 ..     29 ]         17            0
[     30 ..     59 ]         45            1
[     60 ..     92 ]         76            2
[     93 ..    129 ]        111            3
[    130 ..    169 ]        150            4
[    170 ..    213 ]        192            5
[    214 ..    260 ]        237            6
[    261 ..    310 ]        286            7
[    311 ..    364 ]        338            8
[    365 ..    421 ]        393            9
[    422 ..    482 ]        452           10
[    483 ..    546 ]        515           11
[    547 ..    614 ]        581           12
[    615 ..    685 ]        650           13
[    686 ..    759 ]        723           14
[    760 ..    837 ]        799           15
[    838 ..    918 ]        878           16
[    919 ..   1003 ]        961           17
[   1004 ..   1091 ]       1048           18
[   1092 ..   1182 ]       1137           19
[   1183 ..   1277 ]       1230           20
[   1278 ..   1375 ]       1327           21
[   1376 ..   1477 ]       1427           22
[   1478 ..   1582 ]       1530           23
[   1583 ..   1690 ]       1637           24
[   1691 ..   1802 ]       1747           25
[   1803 ..   1917 ]       1860           26
[   1918 ..   2036 ]       1977           27
[   2037 ..   2158 ]       2098           28
[   2159 ..   2283 ]       2221           29
[   2284 ..   2412 ]       2348           30
[   2413 ..   2544 ]       2479           31
[   2545 ..   2680 ]       2613           32
[   2681 ..   2819 ]       2750           33
[   2820 ..   2961 ]       2891           34
[   2962 ..   3107 ]       3035           35
[   3108 ..   3256 ]       3182           36
[   3257 ..   3409 ]       3333           37
[   3410 ..   3565 ]       3488           38
[   3566 ..   3724 ]       3645           39
[   3725 ..   3887 ]       3806           40
[   3888 ..   4053 ]       3971           41
[   4054 ..   4095 ]       4095           42

No infile, done.


>hbp -p_max_levels 4095 -p_multn 0 -p_poissn 0.441128967 -p_addn 44.10389456 -p_darklev 17.83913285 -f_in mp_fib_001.pgm -f_out HB.pgm

output tables as above, perform Histogram Binning on mp_fib_001.pgm, save resulting image as HB.pgm (should have the same contents as mp_fib_001_hb.pgm) 


>hbp -p_max_levels 4095 -p_multn 0 -p_poissn 0.441128967 -p_addn 44.10389456 -p_darklev 17.83913285 -f_in mp_fib_001.pgm -f_out HBP.pgm -raw_to_HBP

output tables as above, perform Histogram Binning and Packing on mp_fib_001.pgm, save resulting image as HBP.pgm (should have the same contents as mp_fib_001_hbp.pgm) 


>hbp -p_max_levels 4095 -p_multn 0 -p_poissn 0.441128967 -p_addn 44.10389456 -p_darklev 17.83913285 -f_in HBP.pgm -f_out re-HB.pgm -HBP_to_HB

output tables as above, recreate Histogram Binned image re-HB.pgm from Histogram Binned and Packed image HBP.pgm (re-HB.pgm should be the same as HB.pgm). Note that for HBP to HB conversion, aside the HBP image, You have to store the noise parameters, or the histogram of HB image