查看: 2693|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
$ m9 S% A* _. C+ p\" N# -*- coding: utf-8 -*-
Q4 `\" A\" g w0 ]8 x$ U1 R
( Z) T1 d+ w& ~: n }1 I% wimport base64
5 K! e& ~% O% K: Afrom os.path import exists; f$ u- j: @) R _1 x7 D/ R8 d\" ^
from tkinter import Tk' ~* w* @+ d0 J- J1 H: `' B
from tkinter.filedialog import askopenfilename0 v( C7 Q* ?0 c& `( q
from urllib.parse import urlencode8 g- G1 x; E7 \* Y: W: Z; C, B3 j
, c: x& A0 |( X3 c. J: Mimport requests1 k A0 z( J# b0 Y' d* N
* I\" N/ `, C# d2 @; B8 u9 {def Dialog_box():. Q7 M# ~' b9 O) b9 ?# G\" M' _/ v
root = Tk()
; M4 N$ }- L* _/ M root.withdraw()
& ]$ j7 I6 C; e% h( q% R F0 e9 X- r% _
filetypes = [(
( ~* `$ G5 ^8 S '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
0 g8 Y& {: F* Q3 m( _) `% O( l ('所有文件(*.*)', '*.*')]3 G0 t9 B- m; Y! J
- ^4 Q: U/ K; v. i, t8 c2 z- s file_path = askopenfilename(filetypes=filetypes)
- J6 N3 @; W& W- _ return file_path
- c2 K* q$ W9 N& v0 t
7 L& l2 z4 ], Y ^' {: U$ L
3 h6 U s }- L0 E# i\" ?def Read_file(file_path):
, r. L, ?\" L4 s) S( b4 R with open(file_path, mode='rb') as file_object:
8 n' Z* C* h1 d7 f/ v' k2 g; B contents = file_object.read()1 M0 R8 y3 T9 ]+ i% Q) F
return contents.rstrip()! g$ ~\" M2 v: t) N
1 R% S( ^8 P0 ]+ Q+ k2 M
$ Z9 d; }& K, L$ F# o
def Baidu_ocr(binary):5 g; x: [- O4 P+ k; x3 J$ L
url = "https://aip.baidubce.com/oauth/2.0/token?"
( U! B* _. J0 f+ l% ] data = {
. d7 {1 t- N% W\" Q+ h5 I "grant_type": "client_credentials",
( J7 ^; A4 M- z: I& M0 r "client_id": "", # 必填
+ f/ ~+ w9 @& U* u% Z8 q "client_secret": "" # 必填- V# M0 g+ G8 O/ K
}+ d4 k5 j/ K1 N2 |! U7 b7 s- F
url += urlencode(data)
7 b D) U* `4 c5 `. A response = requests.get(url=url).json()
2 L) t# B5 Y' Q& t6 k access_token = response['access_token']) e, \' ]: d/ T8 R+ [& Z
& n: t+ T; q1 N9 y$ v4 z4 O0 ~- a% \
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token( g9 z. V7 @; c
post_data = {5 d: t) q0 Y# r6 R, ]( u& K- @
"image": base64.b64encode(binary),
+ Y6 \\" \7 c\" m3 F# X "url": "",
! u& o1 k: b! i* J! C: @ "language_type": "CHN_ENG",; v! l; V, J, T, x+ G# `
"detect_direction": "false",- r9 P9 b% o0 t, z
"detect_language": "false",
( [% H+ a, i7 I, \* n "probability": "false"$ h) p1 E/ Y& v7 r* L
}
' g# R1 I8 Y( m7 t response = requests.post(url=url, data=post_data).json()
4 N% U5 Q9 d6 }! P num = response['words_result_num']' E& q\" t0 ?6 u3 Q; z8 T
result = ''
- B' D1 D6 w e* W+ z6 @& S for i in range(num):
4 Q' N( `# K1 m result += response['words_result'][ i]['words'] + "\r\n"
5 ^$ _# [* e0 G \ return result5 b) `0 v0 ]3 w/ p9 b* _
# ^3 ]2 n9 A ?! u* F( K7 s/ W# u5 ?$ G\" z1 v3 \. t4 s
if __name__ == "__main__":: z0 \( X\" o/ \0 E
file_path = Dialog_box()8 X; o- r% L# u& X1 l! y
if exists(file_path):
/ }( L9 |2 T1 v% ]. `/ Y6 w data = Read_file(file_path): T0 F0 T% |2 B6 h; Y
print(Baidu_ocr(data))# h; S, s0 T; C7 p' K
5 M+ _6 h0 M/ c\" Y: t

#!/usr/bin/python 
$ m9 S% A* _. C+ p\" N# -*- coding: utf-8 -*- 
  Q4 `\" A\" g  w0 ]8 x$ U1 R 
( Z) T1 d+ w& ~: n  }1 I% wimport base64 
5 K! e& ~% O% K: Afrom os.path import exists; f$ u- j: @) R  _1 x7 D/ R8 d\" ^ 
from tkinter import Tk' ~* w* @+ d0 J- J1 H: `' B 
from tkinter.filedialog import askopenfilename0 v( C7 Q* ?0 c& `( q 
from urllib.parse import urlencode8 g- G1 x; E7 \* Y: W: Z; C, B3 j 
 
, c: x& A0 |( X3 c. J: Mimport requests1 k  A0 z( J# b0 Y' d* N 
 
$ n; g5 Z7 e3 G\" v  w# e6 ]# E\" ?6 T3 p 
* I\" N/ `, C# d2 @; B8 u9 {def Dialog_box():. Q7 M# ~' b9 O) b9 ?# G\" M' _/ v 
    root = Tk() 
; M4 N$ }- L* _/ M    root.withdraw() 
& ]$ j7 I6 C; e% h( q% R  F0 e9 X- r% _ 
    filetypes = [( 
( ~* `$ G5 ^8 S        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
0 g8 Y& {: F* Q3 m( _) `% O( l        ('所有文件(*.*)', '*.*')]3 G0 t9 B- m; Y! J 
 
- ^4 Q: U/ K; v. i, t8 c2 z- s    file_path = askopenfilename(filetypes=filetypes) 
- J6 N3 @; W& W- _    return file_path 
- c2 K* q$ W9 N& v0 t 
7 L& l2 z4 ], Y  ^' {: U$ L 
3 h6 U  s  }- L0 E# i\" ?def Read_file(file_path): 
, r. L, ?\" L4 s) S( b4 R    with open(file_path, mode='rb') as file_object: 
8 n' Z* C* h1 d7 f/ v' k2 g; B        contents = file_object.read()1 M0 R8 y3 T9 ]+ i% Q) F 
        return contents.rstrip()! g$ ~\" M2 v: t) N 
1 R% S( ^8 P0 ]+ Q+ k2 M 
$ Z9 d; }& K, L$ F# o 
def Baidu_ocr(binary):5 g; x: [- O4 P+ k; x3 J$ L 
    url = "https://aip.baidubce.com/oauth/2.0/token?" 
( U! B* _. J0 f+ l% ]    data = { 
. d7 {1 t- N% W\" Q+ h5 I        "grant_type": "client_credentials", 
( J7 ^; A4 M- z: I& M0 r        "client_id": "",  # 必填 
+ f/ ~+ w9 @& U* u% Z8 q        "client_secret": ""  # 必填- V# M0 g+ G8 O/ K 
    }+ d4 k5 j/ K1 N2 |! U7 b7 s- F 
    url += urlencode(data) 
7 b  D) U* `4 c5 `. A    response = requests.get(url=url).json() 
2 L) t# B5 Y' Q& t6 k    access_token = response['access_token']) e, \' ]: d/ T8 R+ [& Z 
& n: t+ T; q1 N9 y$ v4 z4 O0 ~- a% \ 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token( g9 z. V7 @; c 
    post_data = {5 d: t) q0 Y# r6 R, ]( u& K- @ 
        "image": base64.b64encode(binary), 
+ Y6 \\" \7 c\" m3 F# X        "url": "", 
! u& o1 k: b! i* J! C: @        "language_type": "CHN_ENG",; v! l; V, J, T, x+ G# ` 
        "detect_direction": "false",- r9 P9 b% o0 t, z 
        "detect_language": "false", 
( [% H+ a, i7 I, \* n        "probability": "false"$ h) p1 E/ Y& v7 r* L 
    } 
' g# R1 I8 Y( m7 t    response = requests.post(url=url, data=post_data).json() 
4 N% U5 Q9 d6 }! P    num = response['words_result_num']' E& q\" t0 ?6 u3 Q; z8 T 
    result = '' 
- B' D1 D6 w  e* W+ z6 @& S    for i in range(num): 
4 Q' N( `# K1 m        result += response['words_result'][ i]['words'] + "\r\n" 
5 ^$ _# [* e0 G  \    return result5 b) `0 v0 ]3 w/ p9 b* _ 
 
# ^3 ]2 n9 A  ?! u* F( K7 s/ W# u5 ?$ G\" z1 v3 \. t4 s 
if __name__ == "__main__":: z0 \( X\" o/ \0 E 
    file_path = Dialog_box()8 X; o- r% L# u& X1 l! y 
    if exists(file_path): 
/ }( L9 |2 T1 v% ]. `/ Y6 w        data = Read_file(file_path): T0 F0 T% |2 B6 h; Y 
        print(Baidu_ocr(data))# h; S, s0 T; C7 p' K 
5 M+ _6 h0 M/ c\" Y: t

转发自派生社区

Python交流群：1047602540

zan