############### HEADER PROFILES #########################################
### First we define header profiles.  Header profiles are not intended
### to be used directly, but instead are suppose to be inherited by other
### profiles.


##################
### RESOLUTION ###
##################
## These header profile define how the scaling of the mask, the foreground
## and the background occur as a function of the resolution.
## They have been designed for SCANNED documents.
##
## We also set here resolution dependent parameters for the segmenter.
## render-size: any character with #render_size# or less pixels will be
## removed. 
##
## edge-size: Defines the thickness (in pixel) of character edges, where
##   the color is unreliable
## blurring-size: size (in pixel) of blurring convolutions applied in the
##   segmenter.
##
## These parameters have been tuned only in the case of normal quality
## scanned document, and they are the default parameters.
##
## For non-scanned document, or scanned document requiring fast or
## aggressive compression, we prefer resolution independent parameters
## for two reasons:
##   1. Segmentation is easier, and parameters are less critical
##   2. It would represent too much testing and tuning time.

# For 300 - 600 DPI scaling there is no need to upsample the mask

### 600 DPI
##
H-600dpi:
dpi=-600
# Keep the image at 600 DPI
upsample=1
subsample=1
# Keep the mask at 600 DPI
mask-upsample=1
# Reduce the background to 100 DPI
bg-subsample=6
# Reduce the foreground to 50 DPI
fg-subsample=12
# Primarily view at 100 DPI
target-subsample=6
resolution-multiplier=2
render-size= 5
edge-size=3
blurring-size=3
high-variation-foreground=false
thickening=1

### 500 DPI
##
H-500dpi:
dpi=-500
# Keep the image at 500 DPI
upsample=1
subsample=1
# Keep the mask at 500 DPI
mask-upsample=1
# Reduce the background to 100 DPI
bg-subsample=5
# Reduce the foreground to 42 DPI
fg-subsample=12
# Primarily view at 100 DPI
target-subsample=5
resolution-multiplier=2
# Minimum object size in pixels in the 500 DPI mask
render-size= 5
edge-size=3
blurring-size=3
high-variation-foreground=false
thickening=1


### 400 DPI
##
H-400dpi:
dpi=-400
# Keep the image at 400 DPI
upsample=1
subsample=1
# Keep the mask at 400 DPI
mask-upsample=1
# Reduce the background to 100 DPI
bg-subsample=4
# Reduce the foreground to 33 DPI
fg-subsample=12
# Primarily view at 100 DPI
target-subsample=4
resolution-multiplier=1
# Minimum object size in pixels in the 400 DPI mask
render-size= 4
edge-size=3
blurring-size=3
high-variation-foreground=false
thickening=1


### 300 DPI
##
H-300dpi:
dpi=-300
# Keep the image at 300 DPI
upsample=1
subsample=1
# Keep the mask at 300 DPI
mask-upsample=1
# Reduce the background to 100 DPI
bg-subsample=3
# Reduce the foreground to 25 DPI
fg-subsample=12
# Primarily view at 100 DPI
target-subsample=3
resolution-multiplier=1
# Minimum object size in pixels in the 300 DPI mask
render-size= 3
# edge-size=3 gives soother and slightly smaller background
# on most documents, but completely removes the background color in some cases.
edge-size=2
blurring-size=3
high-variation-foreground=false
thickening=1


### 300 DPI with an upsampled mask, only for special cases
##
H-300dpi-up:
dpi=-300
# Resize the image to 200 DPI
upsample=2
subsample=3
# Make the mask 400 DPI
mask-upsample=2
# Reduce the background to 100 DPI
bg-subsample=2
# Reduce the foreground to 33   DPI
# Cannot go lower because we must have mask-upsample*fg-subsample<=12
fg-subsample=6 
# Primarily view at 100 DPI
target-subsample=2
# Minimum object size in pixels in the 300 DPI mask
render-size= 3
edge-size=3
blurring-size=3
# the reason this profile is used is for very high sensitivity to text
high-variation-foreground= true
thickening=1


## Full upsampling
## NOT USED OR SUPPORTED
H-300dpi-up-full:
dpi=-300
# Resize the image to 300 DPI
upsample=1
subsample=1
# Make the mask 600 DPI
mask-upsample=2
# Reduce the background to 100 DPI
bg-subsample=3
# Reduce the foreground to 50 DPI
# Cannot go lower because we must have mask-upsample*fg-subsample<=12
fg-subsample=6
# Primarily view at 100 DPI
target-subsample=3
# Minimum object size in pixels in the 300 DPI mask
render-size= 3
edge-size=3
blurring-size=3
high-variation-foreground= false
thickening=1


## 200 DPI with no upsampling on the mask
##
H-200dpi:
dpi=-200
# Keep the image at 200 DPI
upsample=1
subsample=1
# Keep the mask at 200 DPI
mask-upsample=1
# Reduce the background to 100 DPI
bg-subsample=2
# Reduce the foreground to 25 DPI
fg-subsample=8
# Primarily view at 100 DPI
target-subsample=2
resolution-multiplier=1
# Minimum object size in pixels in the 200 DPI mask
render-size= 2
edge-size=2
blurring-size=2
high-variation-foreground=true
thickening=0



### 200 DPI with an upsampled mask
##
H-200dpi-up:
dpi=-200
# Keep the image at 200 DPI
upsample=1
subsample=1
# Reduce the background to 100 DPI
bg-subsample=2
# Primarily view at 100 DPI
target-subsample=2
# Set the mask at 400 DPI
mask-upsample=2
# Reduce the foreground to 33 DPI
fg-subsample=6
# Minimum object size in pixels in the 400 DPI mask
render-size= 4
edge-size=3
blurring-size=3
high-variation-foreground=false
thickening=1


### 150 DPI with no upsampling on the mask: should only be used for
### clean scans
##
H-150dpi:
dpi=-150
# Keep the image at 150 DPI
upsample=1
subsample=1
# Keep the mask at 150 DPI
mask-upsample=1
# Reduce the background to 150 DPI
bg-subsample=1
# Reduce the foreground to 25 DPI
fg-subsample=6
# Primarily view at 150 DPI
target-subsample=1
resolution-multiplier=1
# Minimum object size in pixels in the 150 DPI mask
render-size= 2
edge-size=2
blurring-size=2
high-variation-foreground= true
# JB2 clustering must be more conservative
conservative= true
tolerance-percent=1
thickening=0


### 150 DPI with an upsampled mask
##
H-150dpi-up:
dpi=-150
# Resize the image to 100 DPI
upsample=2
subsample=3
# Make the mask 300 DPI
mask-upsample=3
# Reduce the background to 100 DPI
bg-subsample=1
# Reduce the foreground to 25 DPI
fg-subsample=4
# Primarily view at 100 DPI
target-subsample=1
# Minimum object size in pixels in the 300 DPI mask
render-size= 3
edge-size=3
blurring-size=3
high-variation-foreground= false
thickening=1


### 100 DPI with no upsampling on the mask: should only be used for clean scans
##
H-100dpi:
dpi=-100
# Keep the image at 100 DPI
upsample=1
subsample=1
# Make the mask 100 DPI
mask-upsample=1
# Reduce the background to 100 DPI
bg-subsample=1
# Reduce the foreground to 25 DPI
fg-subsample=4
# Primarily view at 100 DPI
target-subsample=1
resolution-multiplier=1
# Minimum object size in pixels in the 100 DPI mask
render-size= 1
edge-size=1
blurring-size=1
high-variation-foreground= true
# JB2 clustering must be more conservative
conservative= true
tolerance-percent=1
thickening=0

### 100 DPI with an upsampled mask
##
H-100dpi-up:
dpi=-100
# Keep the image at 100 DPI
upsample=1
subsample=1
# Reduce the background to 100 DPI
bg-subsample=1
# Reduce the foreground to 25 DPI
fg-subsample=4
# Primarily view at 100 DPI
target-subsample=1
# Make the mask 300 DPI
mask-upsample=3
# Minimum object size in pixels in the 300 DPI mask
render-size=3
edge-size=3
blurring-size=3
high-variation-foreground= false
thickening=1



###############
### QUALITY ###
###############

### Normal quality compression
##

H-normal-quality:
# We pick a high default quality of 75, this is appropriate for sources
# equivalent to 95 quality jpeg.
quality=75
# Color JB2 increases the compression ratio, but does not allow two
# colors to touch each other in the foreground.  So we do not use
# color JB2 for normal compression.
color-jb2=false
# background-floss= true avoids artifacts
# but make files 3% larger on average, and on dithered documents,
# result into an irregular background.
background-floss=true
subsample-refine=false
limit-mem-usage=false

### Clean compression (Assumes the document is not scanned )

H-clean-quality:
# this has not been tuned
quality=75
color-jb2=false
background-floss=false
subsample-refine=false
limit-mem-usage=false
# Minimum object size in pixels in always 1 pixel
render-size= 1
edge-size=1
blurring-size=1


### Aggressive compression (Lower quality, but smaller.)
##  (THIS HEADER PROFILE IS NOT SUPPORTED!)

H-aggressive-quality:
# The difference between 70 and 75 % quality is barely noticeable in
# most documents visually, but it does significantly reduce the
# compression ratio
quality=70
# Color JB2 increases the compression ratio, but does not allow two
# colors to touch each other in the foreground.
color-jb2=true
background-floss=false
subsample-refine=false
limit-mem-usage=false
edge-size= 3

### Fast compression (Lower quality, but faster, and using less memory.)
##  (THIS HEADER PROFILE IS NOT SUPPORTED!)

H-fast-quality:
color-jb2=false
background-floss=false
subsample-refine=false
limit-mem-usage=true
## lowering blurring-size turns off time-consuming half-toning and texture
## detection algorithms; however, filter gets much slower.
# blurring-size= 2





#######################
### SOURCE DOCUMENT ###
#######################

### Normal Size Text
#  This should be fine for documents with 6 point fonts or larger
#  at 300 DPI.
H-normal-text:

# Take a pixel whose luminance is between the background average and
# the foreground average.  Its relative luminance is a number between
# 00 (background) and 100 (foreground).
#
# This Pixel is classified as foreground only if its relative 
# luminance is larger than threshold-level.
threshold-level=75

# must be conservative here, other characters are lost on 200dpi
# images, or documents with small low-contrast fonts.
pix-filter-level=25

# this parameters has very little influence on most images,
# but may be very dramatic on few
inhibit-foreback-level=40

# We are biased towards dark letters on a white field
inversion-level= 25

# the more aggressive "shape-filter-level= 50" loses too many dots over 'i',
# and sometimes small characters
shape-filter-level= 40
block-overlap= 1





### Drawings, drawn Text, handwriting
##
H-drawn: H-normal-text
inversion-level= 0
pages-per-dict=1
block-overlap= 2
# This option was considered, but is mostly detrimental
# high-variation-foreground= true

### Manuscripts are documents with no photos, and paper that is showing age.
##

H-manuscript: H-drawn

# The ink may  have become very pale, so the segmenter must be more
# sensitive. As there are no photos, ugly artifacts due to
# over-segmentation are less likely.
threshold-level= 60
pix-filter-level= 20
shape-filter-level= 25
background-floss=false

### Maps.
## (THIS PROFILE IS NOT SUPPORTED)
#  Letters are on top of background with many irregular shapes and lines
#  Maps inherit from "test" rather then "drawn" because
#  they use a lot of printed text, which is often inverted.
H-map: H-normal-text
pix-filter-level= 15
shape-filter-level= 25
inversion-level= 10
# high-variation-foreground= true


############### USER PROFILES ###########################################
### Now we define the profiles that are intended for user use.  Most of
### these can be created by joining the header profiles together...

################
### SCANNED  ###
################

#################################################################
# general profile for documents which are scanned 
# they follow the following characteristics
# - text may be inverted.
# - JB2 matching helps
# - foreground color reasonably constant

scan600: H-600dpi H-normal-quality H-normal-text
description="Scanned 600 DPI Documents"
pix-filter-level= 20
## To reduce the file size: this choice must be tested on more images
shape-filter-level= 50

scan500: H-500dpi H-normal-quality H-normal-text
description="Scanned 500 DPI Documents"

scan400: H-400dpi H-normal-quality H-normal-text
description="Scanned 400 DPI Documents"

scan300: H-300dpi H-normal-quality H-normal-text
description="Scanned 300 DPI Documents"

## H-200dpi-up looks better than H-200dpi, at the expense of file size
scan200: H-200dpi-up H-normal-quality H-normal-text
description="Scanned 200 DPI Documents"

scan150: H-150dpi-up H-normal-quality H-normal-text
description="Scanned 150 DPI Documents"

scan100: H-100dpi-up H-normal-quality H-normal-text
description="Scanned 100 DPI Documents"


#############################################################################
# general profile for documents which are drawn rather than machine-printed
# they follow the following characteristics
# - ink is always darker than background: no inversion
# - JB2 matching of little help: no multi-page
# - foreground color may vary a lot: high-variation-foreground
#
# Examples of documents which are drawn
# - comics
# - manuscript
# - blue prints and architectural sketches
# - circuit design
# - sheet music

#############
### DRAWN ###
#############

### Draw document profile
## Author: P. Haffner
##
draw600: H-600dpi H-normal-quality H-drawn
description="Drawn Scanned 600DPI Documents"

draw500: H-500dpi H-normal-quality H-drawn
description="Drawn Scanned 500DPI Documents"

draw400: H-400dpi H-normal-quality H-drawn
description="Drawn Scanned 400DPI Documents"

draw300: H-300dpi H-normal-quality H-drawn
description="Drawn Scanned 300DPI Documents"

## H-200dpi-up looks better than H-200dpi, at the expense of file size
draw200: H-200dpi-up H-normal-quality H-drawn
description="Drawn Scanned 200DPI Documents"

draw150: H-150dpi-up H-normal-quality H-drawn
description="Drawn Scanned 150DPI Documents"

draw100: H-100dpi-up H-normal-quality H-drawn
description="Drawn Scanned 100DPI Documents"


###################
### MANUSCRIPTS ###
###################

### manuscript document profile. (300 DPI)
### Draw document profile
## Author: P. Haffner
##
manuscript600: H-600dpi H-normal-quality H-manuscript
description="Manuscript Scanned 600DPI Documents"

manuscript500: H-500dpi H-normal-quality H-manuscript
description="Manuscript Scanned 500DPI Documents"

manuscript400: H-400dpi H-normal-quality H-manuscript
description="Manuscript Scanned 400DPI Documents"

manuscript300: H-300dpi H-normal-quality H-manuscript
description="Manuscript Scanned 300DPI Documents"

## H-200dpi-up looks better than H-200dpi, at the expense of file size
manuscript200: H-200dpi-up H-normal-quality H-manuscript
description="Manuscript Scanned 200DPI Documents"

manuscript150: H-150dpi-up H-normal-quality H-manuscript
description="Manuscript Scanned 150DPI Documents"

manuscript100: H-100dpi-up H-normal-quality H-manuscript
description="Manuscript Scanned 100DPI Documents"

###############################################################################
###############################################################################
######  EVERYTHING PAST THIS POINT IS UNSUPPORTED AND SHOULD CONSIDERED  ######
######  AS EXPERIMENTAL. ######################################################
###############################################################################

###############################################################################
## general profile for an aggressive treatment of scanned documents ###########
###############################################################################
## THESE PROFILES HAVE BEEN TESTED FOR SMALL SIZE RATHER THAN QUALITY #########
###############################################################################
## The compressed file is generally smaller.
## Author: P. Haffner
##
aggressive600: H-600dpi H-aggressive-quality H-normal-text
description="Scanned 600 DPI Documents"
pix-filter-level= 20
## Reduce the file size...
shape-filter-level= 50

aggressive500: H-500dpi H-aggressive-quality H-normal-text
description="Aggressive compression for Scanned 500 DPI Documents"

aggressive400: H-400dpi H-aggressive-quality H-normal-text
description="Aggressive compression for Scanned 400 DPI Documents"

aggressive300: H-300dpi H-aggressive-quality H-normal-text
description="Aggressive compression for Scanned 300 DPI Documents"

## H-200dpi-up looks better than H-200dpi, but we want to be aggressive!
aggressive200: H-200dpi H-aggressive-quality H-normal-text
description="Aggressive compression for Scanned 200 DPI Documents"

aggressive150: H-150dpi-up H-aggressive-quality H-normal-text
description="Aggressive compression for Scanned 150 DPI Documents"

aggressive100: H-100dpi-up H-aggressive-quality H-normal-text
description="Aggressive compression for Scanned 100 DPI Documents"

very-aggressive300: aggressive300
description="Very Aggressive Compression for Scanned Documents"
shape-filter-level= 60
pix-filter-level= 50
bg-subsample=4
aggressive= true

fast300: H-300dpi H-fast-quality H-normal-text
description="Fast compression for Scanned 300 DPI Documents"

###############################################################################
###############################################################################
######  RECOVERY PROFILES:                                                  ###
######  Used to improve documents where the scan300 profiles misses         ###
######  characters in most cases, these profiles are based on very          ###
######  few  examples, and probably do not match your documents.            ###
###############################################################################
######  Only 300dpi so far, as a recovery profile should only be designed   ###
######  after the profile for a given resolution has been fully tuned.      ###
###############################################################################



################# RECOVER  FINE TEXT ################################
## WARNING: file will be significantly larger than scan300

fine-text300: H-300dpi-up H-normal-quality H-normal-text
description="Scanned 300 DPI Documents With Hard to Segment Text"
threshold-level= 50

################# RECOVER INVERTED TEXT ################################
## only use when all the document is inverted
## This profile works better than "H-normal-text" when most of the text
## is inverted (white letters on a darker background).

invert300: scan300
description="Scanned 300 DPI Documents With Inverted Text"
inversion-level= 75

################# RECOVER  FINE INVERTED TEXT ################################
## WARNING: file will be significantly larger than invert300

fine-invert300: fine-text300
description="Scanned 300 DPI Documents With Hard to Segment Inverted Text"
inversion-level= 75



################# MAP PROFILES #####################################

map600: H-600dpi H-normal-quality H-map
description="Scanned 600 DPI Maps"

map500: H-500dpi H-normal-quality H-map
description="Scanned 500 DPI Maps"

map400: H-400dpi H-normal-quality H-map
description="Scanned 400 DPI Maps"

map300: H-300dpi H-normal-quality H-map
description="Scanned 300 DPI Maps"

map200: H-200dpi H-normal-quality H-map
description="Scanned 200 DPI Maps"

map150: H-150dpi-up H-normal-quality H-map
description="Scanned 150 DPI Maps"

map100: H-100dpi-up H-normal-quality H-map
description="Scanned 100 DPI Maps"
edge-size= 1
blurring-size= 1


##############
### CLEAN  ###
##############

##################################################################
# general profile for documents which are electronically produced.
# They need the following characteristics.
# - JB2 matching helps
# - text may be inverted.
# - foreground color reasonably constant
# - The background is very constant in regions with text.
# - The image has not been significantly reduced in quality
#   with filters, such as saving low as low quality jpeg.
# Postscript produced directly from word processing documents,
# screen dumps, and web page postscript dumps are good candidates
# for this profile.

clean600: H-600dpi H-normal-quality H-normal-text
description="Clean (not scanned) 600 DPI Documents"
pix-filter-level= 20
threshold-level=50
inhibit-foreback-level= 0
resolution-multiplier=1

clean500: H-500dpi H-clean-quality H-normal-text
description="Clean (not scanned) 500 DPI Documents"
pix-filter-level= 20
threshold-level=50
inhibit-foreback-level= 0
resolution-multiplier=1

clean400: H-400dpi H-clean-quality H-normal-text
description="Clean (not scanned) 400 DPI Documents"
pix-filter-level= 20
threshold-level=50
inhibit-foreback-level= 0

clean300: H-300dpi H-clean-quality H-normal-text
description="Clean (not scanned) 300 DPI Documents"
pix-filter-level= 20
threshold-level=50
inhibit-foreback-level= 0

clean200: H-200dpi H-clean-quality H-normal-text
description="Clean (not scanned) 200 DPI Documents"
pix-filter-level= 20
threshold-level=50
inhibit-foreback-level= 0

clean150: H-150dpi H-clean-quality H-normal-text
description="Clean (not scanned) 150 DPI Documents"
pix-filter-level= 20
threshold-level=50
inhibit-foreback-level= 0

clean100: H-100dpi H-clean-quality H-normal-text
description="Clean (not scanned) 100 DPI Documents"
pix-filter-level= 20
threshold-level=50
inhibit-foreback-level= 0



##############################################################
## BITONAL PROFILES


### Bitonal (it is preferable to use bitonaltodjvu)
##
bitonal:
tobitonal=true

bitonal600: bitonal
dpi=-600
aggressive=true

bitonal500: bitonal
dpi=-500
aggressive=true

bitonal400: bitonal
dpi=-400

bitonal300: bitonal
dpi=-300

bitonal200: bitonal
dpi=-200

bitonal150: bitonal
dpi=-150
conservative=true

bitonal100: bitonal
dpi=-100
lossless=true

photo:
jb2-format=none

photo600: photo
dpi=-600

photo500: photo
dpi=-500

photo400: photo
dpi=-400

photo300: photo
dpi=-300

photo200: photo
dpi=-200

photo150: photo
dpi=-150

photo100: photo
dpi=-100


### This defines the default profile that will be used when the
## --profile=<name> option is not specified.

### This defines the default profile that will be used when the
## --profile=<name> option is not specified.
documenttodjvu: scan300