-
-
Save MrAch26/5e2aa7e73b508f8ba9133d468efa4348 to your computer and use it in GitHub Desktop.
| from PIL import Image | |
| from scipy.ndimage import gaussian_filter | |
| import numpy | |
| import pytesseract | |
| from PIL import ImageFilter | |
| def solve_captcha(filename): | |
| # thresold1 on the first stage | |
| th1 = 140 | |
| th2 = 140 # threshold after blurring | |
| sig = 1.5 # the blurring sigma | |
| from scipy import ndimage | |
| original = Image.open(filename) | |
| original.save("original.png") # reading the image from the request | |
| black_and_white = original.convert("L") # converting to black and white | |
| black_and_white.save("black_and_white.png") | |
| first_threshold = black_and_white.point(lambda p: p > th1 and 255) | |
| first_threshold.save("first_threshold.png") | |
| blur = numpy.array(first_threshold) # create an image array | |
| blurred = gaussian_filter(blur, sigma=sig) | |
| blurred = Image.fromarray(blurred) | |
| blurred.save("blurred.png") | |
| final = blurred.point(lambda p: p > th2 and 255) | |
| final = final.filter(ImageFilter.EDGE_ENHANCE_MORE) | |
| final = final.filter(ImageFilter.SHARPEN) | |
| final.save("final.png") | |
| number = pytesseract.image_to_string(Image.open('final.png'), lang='eng', | |
| config='--psm 10 --oem 3 -c tessedit_char_whitelist=0123456789').strip() | |
| print("RESULT OF CAPTCHA:") | |
| print(number) | |
| print("===================") | |
| return number |
@arunpurshotam
That’s because you have a setting on line 29 that specifically says : tessedit_char_whitelist=[0123456789](tel:0123456789)') you should change or remove that
i have this image how i can extrect
@AbdulMobinFata
Change line 29 to this :
result = pytesseract.image_to_string(Image.open('final.png'), lang='eng', config='--psm 10 --oem 3')
result = pytesseract.image_to_string(Image.open('final.png'), lang='eng', config='--psm 10 --oem 3' but it does not work
as final png is like this
the final result is "pereud" for this image,can you help on this.
Didn't work for attached image, the result was "5867" for this imagen, can you help on this?
How do I use my own captcha to test this?
@MikeyD-rbg
Can you send more examples of your captcha and with better resolution ?
this snippet is ment for number as mention on line 28 tessedit_char_whitelist=0123456789
@Manedi send more examples
will try
@MikeyD-rbg Can you send more examples of your captcha and with better resolution ?
this snippet is ment for number as mention on line 28
tessedit_char_whitelist=0123456789
I've sent you 10 more examples. unfortunately all the captchas are from a site and all this size.
How many more examples would you like?
I've manged to solve one of the captchas but the z is not capitalized, how do I ensure all thr letters are capitals?
Added this line but still lowercase z;
config='--psm 10 --oem 3 -c tessedit_char_whitelist=0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ'
It gives the result Captured CAPTCHA: 2Y2z7
It has also solved this captcha correctly execpt the lowercase u, however the z is uppercase.
Captured CAPTCHA: 28DuZ
I have got good results with this script. The image preprocessing works really good and is the key to improve the code. Thanks @MrAch26!
@MrAch26 I've now downloaded over 200k captchas 😂 unfortunately all the same resolution, how many should I upload?
Didn't work for attached image
