RUN LENGTH SMOOTHING ALGORITHM(RLSA) is a method mainly used for block segmentation and text discrimination. It helps to extract the necessary block from the document image.
MIT License
- now rlsa function accepts single value/tuple with pair of values. (before we need to call the function twice)
- a single value be assign to both operations.
- a tuple pair of values be assign to horizontal and vertical operations respectively
- package has a method rlsa_fast that can perform Run Length Smoothing in real time for operations where time is a constraint
- pip install pythonRLSA
- python setup.py install
- pip install -r requirements.txt
from pythonRLSA import rlsa
rlsa.rlsa(image_binary, True, True, (10,5)) # passing different values for H and V operations
rlsa.rlsa(image_binary, True, True, [10,5]) # passing different values for H and V operations
rlsa.rlsa(image_binary, True, True, (10)) # passing same value but in tuple
rlsa.rlsa(image_binary, True, True, [10]) # passing same value but in list
rlsa.rlsa(image_binary, True, True, 10) # passing same value as int
rlsa.rlsa(image_binary, True, False, 10.0) # passing same value as float
rlsa.rlsa(image_binary, False, True, 10) # passing same value as int for V operation only
* H - Horizontal
* V - Vertical
- At the end of the readme, ipython snippet is attached
from pythonRLSA.rlsa_fast import rlsa_fast
H_V = rlsa_fast(image_binary, True, True, 10) # performing Horizontal and Vertical operations
H = rlsa_fast(image_binary, True, False, 10) # performing Horizontal operation
V = rlsa_fast(image_binary, False, True, 10) # performing Vertical operation
- '255'(white pixel) wil be converted to '0'(black pixel) in a image, if the number of adjacent 255's are less than the predefined limit "value".
- The "value" varies among the different images.
- value = 3
- input - [0, 0, 255, 255, 255, 0, 0, 255, 0, 0, 255, 0, 255]
- output - [0, 0, 255, 255, 255, 0, 0, 0, 0, 0, 0, 0, 255]
- python pythonRLSA/test_rlsa_unittest.py -v
- python rlsa_fast/test_rlsafast_unittest.py -v
$ test_bool (__main__.TestRLSA) ... ok
$ test_image (__main__.TestRLSA) ... Image must be an numpy ndarray and must be in binary ... ok
$ test_rlsa_hori (__main__.TestRLSA) ... ok
$ test_rlsa_hori_vert (__main__.TestRLSA) ... ok
$ test_rlsa_vert (__main__.TestRLSA) ... ok
$ test_value (__main__.TestRLSA) ... ok
Ran 6 tests in 0.003s
OK
- python3.5+
- Image must be a binary ndarray(255's/1's/0's)
- Must pass a predefined limit, a certain integer "value"
- rlsa
- rlsa_fast
- image - numpy.ndarray(required)
- horizantal - boolean(required)
- vertial - boolean(required)
- value - any positive integer(int)/list/tuple(required)
# convert the image to binary
import cv2
image = cv2.imread('test_images/image.jpg')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
(thresh, image_binary) = cv2.threshold(gray, 150, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)
# function call
from pythonRLSA import rlsa
image_rlsa_horizontal = rlsa.rlsa(image_binary, True, False, 10)
image_rlsa_horizontal_vertical = rlsa.rlsa(image_binary, True, True, [10,5])
Please ensure that you have updated pip to the latest version before installing pythonRLSA.
If you find any bugs/errors in the usage of above code, please raise an issue through Github or send an email to [email protected] with a clear example that can reproduce the issue.