查看: 2811|回复: 0

Python-百度ocr识别

字体大小: 正常放大

檀俾九

7 主题	3 听众	10 积分

升级 5.26%

该用户从未签到

自我介绍: 我本名为我，那就是我

电梯直达

1^#

发表于 2021-1-12 09:38 |只看该作者 |倒序浏览

|招呼Ta 关注Ta

#!/usr/bin/python! ]7 y5 h9 }4 `# B4 O
# -*- coding: utf-8 -*-
8 [: T2 V- W6 H* C1 p1 X
6 U3 @9 j8 o8 A% f* Simport base644 W) g) K7 O% {4 U- p, F( t
from os.path import exists7 q6 d5 U# ?; H. n- [
from tkinter import Tk% B: H& f3 L# L
from tkinter.filedialog import askopenfilename4 Z& \- k7 c) n\" O
from urllib.parse import urlencode( o* c- e9 F) D, x1 X* ~: W
' j4 g- |3 O9 J Pimport requests
( X& {% u- d& B$ X\" n* X) s; q$ m$ Z9 v. X; @
% o: P m9 l$ g# \def Dialog_box():# v4 ^. S1 D& H: }0 o
root = Tk()- H# M$ n8 |. m& L0 V3 w
root.withdraw() o- _& n6 k2 v( L7 H$ F4 K
# v- K( v- n5 u8 L5 c* ?- I filetypes = [(
5 y5 C4 }& R5 V+ X# Q5 o! I '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
& f: g& T2 X. [& d6 g. b) B( } ('所有文件(*.*)', '*.*')]
) F2 l! H8 z% X- v g
5 P$ r5 s! Z) T3 [7 v) z# d' l file_path = askopenfilename(filetypes=filetypes)
# V. U( }. t) q& j. v6 g, i! F, z return file_path) S) ?1 [( e7 Q# F. n2 f5 U( V
; G' Y$ @2 |\" v& }# f A; g; i. Y, n8 a5 r$ l6 O9 y l% f9 O
def Read_file(file_path):
( C: k/ B; S! K' L: V6 D& _4 m with open(file_path, mode='rb') as file_object:5 D& G7 D& g+ {9 I3 N. @7 C
contents = file_object.read(): ], f1 j& T3 r) q' I2 }
return contents.rstrip()
5 M( s8 `8 @% a* `( O- ^/ _8 g8 U+ i; h6 \- N# @+ Q+ a6 h9 q* K
) a+ D, D- c4 x$ Vdef Baidu_ocr(binary):
! g$ K3 v% Z; r* P; N0 T# g1 S) c url = "https://aip.baidubce.com/oauth/2.0/token?"
X, H0 J. `! U0 E7 { A\" a data = {
4 M2 G3 t$ j0 c) ?4 K- ] "grant_type": "client_credentials",
9 u! @/ H [, y: o8 S "client_id": "", # 必填
1 t3 M$ E6 X' n) q+ C\" H: J S "client_secret": "" # 必填2 p2 }4 ? X1 |8 y2 |. G: E
}
( f* O$ m+ y( z$ ^4 m: I2 Y) t url += urlencode(data)- K8 N2 x* _! V
response = requests.get(url=url).json()- M$ I, o. j( P+ O% S3 D8 g5 H2 X) S- Z! T! G
access_token = response['access_token']3 _& q' i. q7 A7 I\" J5 e
, `4 D6 a; d/ \\" v/ k |\" ~
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token% G2 x! O x, G a# J& M4 g7 P
post_data = {9 k! y# \1 i\" j! Q8 A. \% j
"image": base64.b64encode(binary),
( A: I; t\" q) h4 s J "url": "",5 t: I. f8 y) R5 e9 z
"language_type": "CHN_ENG",
; N4 u0 o+ s3 I) y- e "detect_direction": "false",* p t% Y9 }0 F# }' b, r
"detect_language": "false",6 n8 T( {) e# [\" M4 z, ]
"probability": "false"$ V- \/ y7 o- s) S& [, P- i
}+ K$ h+ u; _4 O$ g$ d% d\" K% T
response = requests.post(url=url, data=post_data).json()
1 X: V! F) a0 ` num = response['words_result_num']* `% S3 o- S3 ^1 B3 M. k9 B) a8 c
result = ''
0 P3 O- q& M6 l' X3 P for i in range(num):3 S8 A' c, |+ E. k
result += response['words_result'][ i]['words'] + "\r\n"8 ^( s8 @0 m\" I4 d
return result( I8 w; N# z\" [$ x7 c# \
8 c7 z- d7 z9 V' o
5 X- e' p3 X4 k, C\" ]if __name__ == "__main__":
\" l' N/ j p: M# l w+ { file_path = Dialog_box()* e3 `; v7 o) D
if exists(file_path):9 V\" Y; D5 F6 y. `7 x$ n
data = Read_file(file_path)/ G, o+ J% l9 E, L
print(Baidu_ocr(data))
6 P! P+ e\" t& r, P: G8 q7 ?8 Q/ l T. u/ d' U7 G8 Y

#!/usr/bin/python! ]7 y5 h9 }4 `# B4 O 
# -*- coding: utf-8 -*- 
8 [: T2 V- W6 H* C1 p1 X 
6 U3 @9 j8 o8 A% f* Simport base644 W) g) K7 O% {4 U- p, F( t 
from os.path import exists7 q6 d5 U# ?; H. n- [ 
from tkinter import Tk% B: H& f3 L# L 
from tkinter.filedialog import askopenfilename4 Z& \- k7 c) n\" O 
from urllib.parse import urlencode( o* c- e9 F) D, x1 X* ~: W 
 
' j4 g- |3 O9 J  Pimport requests 
( X& {% u- d& B$ X\" n* X) s; q$ m$ Z9 v. X; @ 
 
% o: P  m9 l$ g# \def Dialog_box():# v4 ^. S1 D& H: }0 o 
    root = Tk()- H# M$ n8 |. m& L0 V3 w 
    root.withdraw()  o- _& n6 k2 v( L7 H$ F4 K 
 
# v- K( v- n5 u8 L5 c* ?- I    filetypes = [( 
5 y5 C4 }& R5 V+ X# Q5 o! I        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
& f: g& T2 X. [& d6 g. b) B( }        ('所有文件(*.*)', '*.*')] 
) F2 l! H8 z% X- v  g 
5 P$ r5 s! Z) T3 [7 v) z# d' l    file_path = askopenfilename(filetypes=filetypes) 
# V. U( }. t) q& j. v6 g, i! F, z    return file_path) S) ?1 [( e7 Q# F. n2 f5 U( V 
 
; G' Y$ @2 |\" v& }# f  A; g; i. Y, n8 a5 r$ l6 O9 y  l% f9 O 
def Read_file(file_path): 
( C: k/ B; S! K' L: V6 D& _4 m    with open(file_path, mode='rb') as file_object:5 D& G7 D& g+ {9 I3 N. @7 C 
        contents = file_object.read(): ], f1 j& T3 r) q' I2 } 
        return contents.rstrip() 
5 M( s8 `8 @% a* `( O- ^/ _8 g8 U+ i; h6 \- N# @+ Q+ a6 h9 q* K 
 
) a+ D, D- c4 x$ Vdef Baidu_ocr(binary): 
! g$ K3 v% Z; r* P; N0 T# g1 S) c    url = "https://aip.baidubce.com/oauth/2.0/token?" 
  X, H0 J. `! U0 E7 {  A\" a    data = { 
4 M2 G3 t$ j0 c) ?4 K- ]        "grant_type": "client_credentials", 
9 u! @/ H  [, y: o8 S        "client_id": "",  # 必填 
1 t3 M$ E6 X' n) q+ C\" H: J  S        "client_secret": ""  # 必填2 p2 }4 ?  X1 |8 y2 |. G: E 
    } 
( f* O$ m+ y( z$ ^4 m: I2 Y) t    url += urlencode(data)- K8 N2 x* _! V 
    response = requests.get(url=url).json()- M$ I, o. j( P+ O% S3 D8 g5 H2 X) S- Z! T! G 
    access_token = response['access_token']3 _& q' i. q7 A7 I\" J5 e 
, `4 D6 a; d/ \\" v/ k  |\" ~ 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token% G2 x! O  x, G  a# J& M4 g7 P 
    post_data = {9 k! y# \1 i\" j! Q8 A. \% j 
        "image": base64.b64encode(binary), 
( A: I; t\" q) h4 s  J        "url": "",5 t: I. f8 y) R5 e9 z 
        "language_type": "CHN_ENG", 
; N4 u0 o+ s3 I) y- e        "detect_direction": "false",* p  t% Y9 }0 F# }' b, r 
        "detect_language": "false",6 n8 T( {) e# [\" M4 z, ] 
        "probability": "false"$ V- \/ y7 o- s) S& [, P- i 
    }+ K$ h+ u; _4 O$ g$ d% d\" K% T 
    response = requests.post(url=url, data=post_data).json() 
1 X: V! F) a0 `    num = response['words_result_num']* `% S3 o- S3 ^1 B3 M. k9 B) a8 c 
    result = '' 
0 P3 O- q& M6 l' X3 P    for i in range(num):3 S8 A' c, |+ E. k 
        result += response['words_result'][ i]['words'] + "\r\n"8 ^( s8 @0 m\" I4 d 
    return result( I8 w; N# z\" [$ x7 c# \ 
 
8 c7 z- d7 z9 V' o 
5 X- e' p3 X4 k, C\" ]if __name__ == "__main__": 
\" l' N/ j  p: M# l  w+ {    file_path = Dialog_box()* e3 `; v7 o) D 
    if exists(file_path):9 V\" Y; D5 F6 y. `7 x$ n 
        data = Read_file(file_path)/ G, o+ J% l9 E, L 
        print(Baidu_ocr(data)) 
6 P! P+ e\" t& r, P: G8 q7 ?8 Q/ l  T. u/ d' U7 G8 Y

转发自派生社区

Python交流群：1047602540

zan