查看: 2845|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
1 m8 f5 x' D1 E# -*- coding: utf-8 -*-
* {# R7 N- A7 R7 W+ f U5 `, ]
# T' V, @- }# o9 h$ {import base64\" j$ Q\" J( U/ p9 Z
from os.path import exists
- w7 ?$ v6 ]; V9 r3 n6 r% g7 D: r+ W+ Tfrom tkinter import Tk
\" {+ P3 r6 ?5 t1 o4 Q% ffrom tkinter.filedialog import askopenfilename
$ A4 `% w) n2 w: ]& Q7 a* hfrom urllib.parse import urlencode
/ H! [! j+ A1 T1 \
& C' A' d1 w. Z2 ?* ?9 Y: e) himport requests% B, Z4 L( b: R `7 T) r9 x
1 R5 q& ^' d, j( |# Y' K
! B- d7 Z8 k% o/ Xdef Dialog_box():1 r8 ~# ~( R* b* F# h1 a\" L
root = Tk(). F\" G\" [9 o _
root.withdraw()
& P% }3 Y4 @6 _+ C) J9 d* j4 ~: f, Y' \; F. H- Y
filetypes = [(0 n0 ]4 s7 M* p+ a0 }2 `! r
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),2 u( {/ A8 p/ f1 _% a8 u
('所有文件(*.*)', '*.*')]
7 J1 X- y\" F# v1 ?- F# @
5 I {2 V$ l\" b6 d% n w file_path = askopenfilename(filetypes=filetypes)
, V5 Z% P0 @\" W2 k- a return file_path
( R( ?+ @# Z6 ~) [# U$ S2 i
% G; x, S* N8 z! k$ W! @+ w9 @% q3 W' _: Y8 i1 H4 e0 T% D
def Read_file(file_path):
5 s, s1 \+ m& _8 N with open(file_path, mode='rb') as file_object:% o: |( S. X( O3 g) x- H
contents = file_object.read()0 I5 }/ M3 `; z% J* D. _
return contents.rstrip() A, t# }5 u5 R+ d& [ f1 }2 K9 w4 `9 v
1 c) M( G6 L- ?! ?9 Y9 R9 Z* e
) o: F q( \0 i, Q/ Odef Baidu_ocr(binary):1 [7 y9 H2 f. K% X% k
url = "https://aip.baidubce.com/oauth/2.0/token?"
, B1 Q3 q! v* Q' Z; K( @ data = {
7 ?\" ]. [0 B9 S6 ]1 X+ s "grant_type": "client_credentials",1 M i' M& e7 u- |! H+ |
"client_id": "", # 必填
- i5 z( n; \' L\" g0 h# h "client_secret": "" # 必填
0 v9 S1 L% R& t9 ?+ _ }
* S& p; w+ ?1 I. B2 A url += urlencode(data). B1 h# R# {% }$ B, n' C2 c: x' Q# l
response = requests.get(url=url).json()
5 q* L5 u. ~, [6 p# p access_token = response['access_token'], \$ x$ K e# P- w5 g0 J1 F
, x. E* w( k' ~8 n1 {
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
l7 Q5 R* \) O. S( s post_data = {- {' M: O3 F* J; R; h+ l2 |7 a* p
"image": base64.b64encode(binary),
( O+ c$ K; I4 W7 n4 ?$ o- S: k, e "url": "",
' k4 b6 x m6 [1 M "language_type": "CHN_ENG",3 ~) X) p/ J1 Y$ B
"detect_direction": "false",+ T% u# ~! ~! X
"detect_language": "false",
& C8 A% S& F1 V, B "probability": "false"9 \2 ]. T/ J# `6 Q. \
}$ t\" u. I& k6 K! i
response = requests.post(url=url, data=post_data).json()5 T0 W1 e3 f) g K. V. A# k* V
num = response['words_result_num']
3 `/ B7 o8 V# P# g+ v result = ''. k3 b0 z4 z* W2 c6 j2 z; }2 K
for i in range(num):& d0 s7 ? C\" ] h, w
result += response['words_result'][ i]['words'] + "\r\n"# m# V# J3 g% Z( D4 c% R2 ~
return result4 D5 T7 e w& Y# ^ l- Q4 z\" A
3 j' \# F: t& E/ o/ w0 q/ D' S
$ g: W- ^% _1 S/ v, Pif __name__ == "__main__":2 B: w# j! c1 n0 G+ S# V
file_path = Dialog_box()
9 o& _6 {1 Q0 B) g4 O, P if exists(file_path):- `9 }+ q- `5 {% ]. ~
data = Read_file(file_path)
( \! f# a2 G+ \! d; E/ j print(Baidu_ocr(data))
# _6 X3 q7 J9 c3 ?; @ u& i6 t\" S |

#!/usr/bin/python 
1 m8 f5 x' D1 E# -*- coding: utf-8 -*- 
* {# R7 N- A7 R7 W+ f  U5 `, ] 
# T' V, @- }# o9 h$ {import base64\" j$ Q\" J( U/ p9 Z 
from os.path import exists 
- w7 ?$ v6 ]; V9 r3 n6 r% g7 D: r+ W+ Tfrom tkinter import Tk 
\" {+ P3 r6 ?5 t1 o4 Q% ffrom tkinter.filedialog import askopenfilename 
$ A4 `% w) n2 w: ]& Q7 a* hfrom urllib.parse import urlencode 
/ H! [! j+ A1 T1 \ 
& C' A' d1 w. Z2 ?* ?9 Y: e) himport requests% B, Z4 L( b: R  `7 T) r9 x 
 
1 R5 q& ^' d, j( |# Y' K 
! B- d7 Z8 k% o/ Xdef Dialog_box():1 r8 ~# ~( R* b* F# h1 a\" L 
    root = Tk(). F\" G\" [9 o  _ 
    root.withdraw() 
& P% }3 Y4 @6 _+ C) J9 d* j4 ~: f, Y' \; F. H- Y 
    filetypes = [(0 n0 ]4 s7 M* p+ a0 }2 `! r 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),2 u( {/ A8 p/ f1 _% a8 u 
        ('所有文件(*.*)', '*.*')] 
7 J1 X- y\" F# v1 ?- F# @ 
5 I  {2 V$ l\" b6 d% n  w    file_path = askopenfilename(filetypes=filetypes) 
, V5 Z% P0 @\" W2 k- a    return file_path 
( R( ?+ @# Z6 ~) [# U$ S2 i 
% G; x, S* N8 z! k$ W! @+ w9 @% q3 W' _: Y8 i1 H4 e0 T% D 
def Read_file(file_path): 
5 s, s1 \+ m& _8 N    with open(file_path, mode='rb') as file_object:% o: |( S. X( O3 g) x- H 
        contents = file_object.read()0 I5 }/ M3 `; z% J* D. _ 
        return contents.rstrip()  A, t# }5 u5 R+ d& [  f1 }2 K9 w4 `9 v 
 
1 c) M( G6 L- ?! ?9 Y9 R9 Z* e 
) o: F  q( \0 i, Q/ Odef Baidu_ocr(binary):1 [7 y9 H2 f. K% X% k 
    url = "https://aip.baidubce.com/oauth/2.0/token?" 
, B1 Q3 q! v* Q' Z; K( @    data = { 
7 ?\" ]. [0 B9 S6 ]1 X+ s        "grant_type": "client_credentials",1 M  i' M& e7 u- |! H+ | 
        "client_id": "",  # 必填 
- i5 z( n; \' L\" g0 h# h        "client_secret": ""  # 必填 
0 v9 S1 L% R& t9 ?+ _    } 
* S& p; w+ ?1 I. B2 A    url += urlencode(data). B1 h# R# {% }$ B, n' C2 c: x' Q# l 
    response = requests.get(url=url).json() 
5 q* L5 u. ~, [6 p# p    access_token = response['access_token'], \$ x$ K  e# P- w5 g0 J1 F 
, x. E* w( k' ~8 n1 { 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
  l7 Q5 R* \) O. S( s    post_data = {- {' M: O3 F* J; R; h+ l2 |7 a* p 
        "image": base64.b64encode(binary), 
( O+ c$ K; I4 W7 n4 ?$ o- S: k, e        "url": "", 
' k4 b6 x  m6 [1 M        "language_type": "CHN_ENG",3 ~) X) p/ J1 Y$ B 
        "detect_direction": "false",+ T% u# ~! ~! X 
        "detect_language": "false", 
& C8 A% S& F1 V, B        "probability": "false"9 \2 ]. T/ J# `6 Q. \ 
    }$ t\" u. I& k6 K! i 
    response = requests.post(url=url, data=post_data).json()5 T0 W1 e3 f) g  K. V. A# k* V 
    num = response['words_result_num'] 
3 `/ B7 o8 V# P# g+ v    result = ''. k3 b0 z4 z* W2 c6 j2 z; }2 K 
    for i in range(num):& d0 s7 ?  C\" ]  h, w 
        result += response['words_result'][ i]['words'] + "\r\n"# m# V# J3 g% Z( D4 c% R2 ~ 
    return result4 D5 T7 e  w& Y# ^  l- Q4 z\" A 
3 j' \# F: t& E/ o/ w0 q/ D' S 
 
$ g: W- ^% _1 S/ v, Pif __name__ == "__main__":2 B: w# j! c1 n0 G+ S# V 
    file_path = Dialog_box() 
9 o& _6 {1 Q0 B) g4 O, P    if exists(file_path):- `9 }+ q- `5 {% ]. ~ 
        data = Read_file(file_path) 
( \! f# a2 G+ \! d; E/ j        print(Baidu_ocr(data)) 
# _6 X3 q7 J9 c3 ?; @  u& i6 t\" S  |

转发自派生社区

Python交流群：1047602540

zan