XML File Format (DTD)
XML File Format (DTD)
The structure of the XML format is
defined in the following
DTD
file:
<!--
This is the DTD for "IAM Handwriting Database" metadata.
The system-id is 'http://www.iam.unibe.ch/~fki/iamDB/form-metadata.dtd', there
is no public-id yet. The java classes map the system-id to the java resource
ch/unibe/iam/handwirtingDB/form-metadata.dtd in the handwirtingDB jar-file.
-->
<!--
Formular
created: creation date, format yyyy-mm-dd
height: height of image data, in pixel
id: identification number of form, format /[a-z]\d\d\d-\d\d\d\w?/
last-modified: modification date, format yyyy-mm-dd
skew: form skew, in 1000th degree
status: form status, is "final"
version: version, is "3.0"
width: width of image data, in pixel
wid: writer identification, a number
-->
<!ELEMENT form (machine-printed-part, handwritten-part)>
<!ATTLIST form
created CDATA #REQUIRED
height CDATA #REQUIRED
id ID #REQUIRED
last-modified CDATA #REQUIRED
skew CDATA #REQUIRED
status (final|verified|segmented|raw) #FIXED "final"
version (3.0_beta|3.0) #REQUIRED
width CDATA #REQUIRED
writer-id CDATA #REQUIRED
select CDATA #IMPLIED
>
<!--
Machine Printed Part
text: content of line, a string
-->
<!ELEMENT machine-printed-part (machine-print-line)+>
<!ELEMENT machine-print-line EMPTY>
<!ATTLIST machine-print-line
text CDATA #REQUIRED
>
<!--
Hand Written Part
-->
<!ELEMENT handwritten-part (line)+>
<!--
Line
ass: ascender slope, 1000th degree
asx: ascender x value, is "0"
asy: ascender y value
character-width: average character width (unused), is "999"
dss: descender slope, 1000th degree
dsx: descender x value, is "0"
dsy: descender y value
fd0: fractal dimension 0 (see master of Caroline Hertel)
fd1: fractal dimension 1
fd2: fractal dimension 2
filter-width: filter width (meaning unknown ???)
id: identifier, format is: indentfier of enclosing
form "-" line number (starting with 0)
lss: lower baseline slope, 1000th degree
lsx: lower baseline x value, is "0"
lsy: lower baseline y value
segmentation: result of word segmentation, is "err" or "ok"
slant: slant, 1000th degree
stroke-width: average stroke width, in pixel
threshold: threshold of binarization
text: content of line, a string
uss: upper baseline slope, 1000th degree
usx: upper baseline x value, is "0"
usy: upper baseline y value
-->
<!ELEMENT line ((word)+, (upper-contour)?, (lower-contour)?)>
<!ATTLIST line
ass CDATA #REQUIRED
asx CDATA #FIXED "0"
asy CDATA #REQUIRED
character-width CDATA #FIXED "999"
dss CDATA #REQUIRED
dsx CDATA #FIXED "0"
dsy CDATA #REQUIRED
fd0 CDATA #IMPLIED
fd1 CDATA #IMPLIED
fd2 CDATA #IMPLIED
filter-width CDATA #REQUIRED
id ID #REQUIRED
lbs CDATA #REQUIRED
lbx CDATA #FIXED "0"
lby CDATA #REQUIRED
segmentation (ok|err) #REQUIRED
slant CDATA #REQUIRED
stroke-width CDATA #REQUIRED
threshold CDATA #REQUIRED
text CDATA #REQUIRED
ubs CDATA #REQUIRED
ubx CDATA #FIXED "0"
uby CDATA #REQUIRED
>
<!-- Word
id: identifier, format is: indentfier of enclosing
line "-" word number (starting with 0)
sentence-start: initial-sentence-marker, is "yes" or "no"
tag: type of word, as defined by ??? standard
text: value of word, a string
-->
<!ELEMENT word (cmp)*>
<!ATTLIST word
id ID #REQUIRED
sentence-start (yes|no) "no"
tag CDATA #REQUIRED
text CDATA #REQUIRED
>
<!-- Component
x: x position
y: y position
width: width, in pixel
height: height, in pixel
-->
<!ELEMENT cmp EMPTY>
<!ATTLIST cmp
x CDATA #REQUIRED
y CDATA #REQUIRED
width CDATA #REQUIRED
height CDATA #REQUIRED
>
<!-- Contour Data
x: x position
y: y position
-->
<!ELEMENT upper-contour (point)+>
<!ELEMENT lower-contour (point)+>
<!ELEMENT point EMPTY>
<!ATTLIST point
x CDATA #REQUIRED
y CDATA #REQUIRED
>
