10 Commits

Author SHA1 Message Date
llocarnini
723c6606e1 kernel size for morphology ex set bit higher, so less non-table structure are detected. Reduced the kernel size of the directional motion blurr and increased the treshold a little bit so narrow cells wont be split up.
Problem with the cell filtering for certain scanned pdfs detected.
2022-02-18 16:35:50 +01:00
llocarnini
d70781f4aa changed tolerance in adjacent1 function in postprocessing.y from 2 to 4
added function so vertical and horizontal components do not overlap the layout box of the table
2022-02-17 16:45:55 +01:00
llocarnini
07907d45dd some changes to fix some minor bugs in table_parsing.py and post_processing.py 2022-02-10 10:56:03 +01:00
llocarnini
4964c8f5a1 some changes to fix some minor bugs in table_parsing.py and post_processing.py 2022-02-10 10:22:22 +01:00
Matthias Bisping
295666c28f added todo comments 2022-02-06 21:25:01 +01:00
Matthias Bisping
90b8613bf8 filtering non-tables by bounding rect check WIP 2022-02-06 21:03:40 +01:00
Matthias Bisping
36284f9a78 removed obsolete lines 2022-02-06 20:01:00 +01:00
Matthias Bisping
0fc6cf8008 fixed bug in adjaceny test 2022-02-06 20:00:38 +01:00
Matthias Bisping
106b333dca filtering for connected cells... but does not quite work yet 2022-02-06 16:44:07 +01:00
Matthias Bisping
e652da1fa8 refactoring 2022-02-06 14:53:17 +01:00