Toolkit for document image processing
Project description
Contextualization
In progress toolkit for document image pre processing.
Aimed for images to be OCRed.
Main Available methods
-
Auto rotate image
Uses left margin of a document to calculate the angle of rotation present, and correct it accordingly.
Can be given the rotation direction (clocwise or counter_clockwise), or in auto mode tries to determine the side to which the document is tilted (can be none, in which case image won't be rotated).
-
Calculate rotation direction
Calculates rotation direction of an image by finding the biggest sets of the first black pixels appearances (with outliers removed) in the image for each direction: clockwise, counter_clockwise and none.
For none direction, the set is created based on pixels with same 'x' coordinate that with less than a 5% height difference, relative to the image's height.
-
Binarize document
-
Split document into columns
Analyzes document image pixel color frequency and split document image into columns.
-
Auto crop document
Analyzes document image pixel color frequency and cut document margins, aiming mostly to remove possible folds in the corners.
-
Identify document images Identify document images in image, using algorithm available in leptonica's repository that finds potential image masks.
-
Get document delimiters Get document delimiters, using image transformations.
-
Segment document Segments document image into header, body and footer, using delimiters. Only the body is always guaranteed to have a value.
Bash commands:
-
binarize : binarize document image.
-
rotate_document : rotate document image.
-
split_columns : split document into column images.
-
d_auto_crop : auto crop document image.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for document_image_utils-0.1.18.7.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | 603a8f458274cf7f8bd7578d2c330c6039e2a54ba85ced19c7b2436abb1b6678 |
|
MD5 | 472577bddf4b3d13db3e703906e13815 |
|
BLAKE2b-256 | 2af6db22057604445982bca34e3065f42dec02ffcad47612c4e66aa30184d42b |
Hashes for document_image_utils-0.1.18.7-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | f58226dc5d2ef7d6440ffbe5f98b7d6e2da75533a66a1269c51b5fa08caf1db0 |
|
MD5 | 7ecbead972330c25540ce6e430e29fee |
|
BLAKE2b-256 | b078d05ba9cccd005583c91f64f2f3033fa6a41dc265b800de2520673b620695 |