查看: 2837|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
9 v: q* t8 v! y1 L& |) l# \# -*- coding: utf-8 -*-( B8 N2 G8 z& t; P. ]\" v0 i
2 ?\" f2 O9 z. ^; h\" himport base64' C8 C) o. z: K8 }5 w3 l
from os.path import exists
D; u3 R6 g. F5 l. ifrom tkinter import Tk
( I! q. z! c( Z, j, X- T/ Nfrom tkinter.filedialog import askopenfilename/ e, z. ?- e d3 E
from urllib.parse import urlencode
) h1 L) t p( a- _* o2 G+ o* @; `9 ~+ c) M% H2 S0 R
import requests
\" g2 A/ ^9 o6 F: ~* n# T
/ V. S P3 b: u- _! i
3 a3 q! L; t- f, l( B: adef Dialog_box():
/ c9 R% p8 U\" \' M/ Y% [( \ root = Tk()
( H% z7 m3 T1 b root.withdraw(): x3 {0 d% @0 o# T6 `
+ d+ K$ E) E7 s* c j6 D7 c
filetypes = [(
7 Z/ w3 D1 ?2 X0 ^ '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),# g2 D/ L. S6 f
('所有文件(*.*)', '*.*')]
+ ~2 z* D( b3 k+ [# K% s1 j. p
0 ]( T, `4 k9 A7 @: n file_path = askopenfilename(filetypes=filetypes)
; `4 c- X$ w) \2 r: ]; j+ g return file_path
5 I, L2 a9 Q. o& \) @) b
y2 d; t4 O6 Bdef Read_file(file_path):
/ y3 W# [$ C- O) C with open(file_path, mode='rb') as file_object:
7 `5 d8 v: J/ P2 m/ w8 N! @ contents = file_object.read()
* ]; ?- ?9 D1 ?( z( _% v0 N return contents.rstrip()6 E5 O7 \: q# p) R2 `/ e
. a2 Q: i( v7 }+ {8 O4 w: Y3 k
7 C3 f' N( p$ V, `
def Baidu_ocr(binary):% A1 \: v, N% J3 o
url = "https://aip.baidubce.com/oauth/2.0/token?"
' j% {: @& J\" p: o' O; v, O data = {* b0 g\" s2 Z) K. L; f8 }
"grant_type": "client_credentials",, c/ s\" t( V! x
"client_id": "", # 必填2 i1 u8 o ?& o3 l* l
"client_secret": "" # 必填
/ L& ^5 I6 {* c, \3 E9 K }
) `) G* a3 e. U4 D5 w( i; O url += urlencode(data)
: q* ?* g n5 I& }* N. P; C7 [8 [ response = requests.get(url=url).json()( r( N* n! R, C
access_token = response['access_token']
! S; X\" H# B* ^
# P; I& ~) W0 y( b5 r url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token U9 Q ~1 u+ o3 ~3 y3 t
post_data = {- k: N% I9 t4 I( w. T\" d
"image": base64.b64encode(binary),# a2 E; `6 P8 ~# U- K' X
"url": "",8 i& c, ^- A1 Y) F5 d4 f+ T+ S) s, R
"language_type": "CHN_ENG",
6 N1 K9 K\" n; b7 i8 o1 @' Y- G) r8 N "detect_direction": "false",
4 u1 [7 z\" x* L& _1 V6 A "detect_language": "false",! q- t\" O$ u% d* f
"probability": "false"% q8 L2 A8 U2 c2 D; `
}' m$ v& Q. q* ]* V: V; }* {9 [
response = requests.post(url=url, data=post_data).json()
1 g3 k# V2 {; m, T3 w [! W( G\" p num = response['words_result_num']: Z( R) T( k\" r$ Y
result = ''' ?1 u. ]8 R# }4 s9 N
for i in range(num):0 N; ~: w& n4 M- }! O. ^
result += response['words_result'][ i]['words'] + "\r\n") k3 f5 F2 o! O, M
return result5 ^0 w) }2 F1 O) F! l2 u
2 o3 Z\" V\" s, F\" \5 Z, s% O
0 _5 ~, \- M. Y! X2 |; wif __name__ == "__main__":5 X3 J8 L S2 Z. N\" g* Y
file_path = Dialog_box()
- n- F3 N\" {8 z if exists(file_path):
5 N9 ^5 r) }/ J* I data = Read_file(file_path)
5 v. b- \# _2 h L- \\" [ w print(Baidu_ocr(data)). W\" b6 S, l$ l2 b2 Q; u( b, n9 H
. g i; A: b( @( ^7 j6 v

#!/usr/bin/python 
9 v: q* t8 v! y1 L& |) l# \# -*- coding: utf-8 -*-( B8 N2 G8 z& t; P. ]\" v0 i 
 
2 ?\" f2 O9 z. ^; h\" himport base64' C8 C) o. z: K8 }5 w3 l 
from os.path import exists 
  D; u3 R6 g. F5 l. ifrom tkinter import Tk 
( I! q. z! c( Z, j, X- T/ Nfrom tkinter.filedialog import askopenfilename/ e, z. ?- e  d3 E 
from urllib.parse import urlencode 
) h1 L) t  p( a- _* o2 G+ o* @; `9 ~+ c) M% H2 S0 R 
import requests 
\" g2 A/ ^9 o6 F: ~* n# T 
/ V. S  P3 b: u- _! i 
3 a3 q! L; t- f, l( B: adef Dialog_box(): 
/ c9 R% p8 U\" \' M/ Y% [( \    root = Tk() 
( H% z7 m3 T1 b    root.withdraw(): x3 {0 d% @0 o# T6 ` 
+ d+ K$ E) E7 s* c  j6 D7 c 
    filetypes = [( 
7 Z/ w3 D1 ?2 X0 ^        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),# g2 D/ L. S6 f 
        ('所有文件(*.*)', '*.*')] 
+ ~2 z* D( b3 k+ [# K% s1 j. p 
0 ]( T, `4 k9 A7 @: n    file_path = askopenfilename(filetypes=filetypes) 
; `4 c- X$ w) \2 r: ]; j+ g    return file_path 
! c7 `2 ]  P! m1 s. a  c6 T; i  f\" D5 I, L2 a9 Q. o& \) @) b 
 
  y2 d; t4 O6 Bdef Read_file(file_path): 
/ y3 W# [$ C- O) C    with open(file_path, mode='rb') as file_object: 
7 `5 d8 v: J/ P2 m/ w8 N! @        contents = file_object.read() 
* ]; ?- ?9 D1 ?( z( _% v0 N        return contents.rstrip()6 E5 O7 \: q# p) R2 `/ e 
. a2 Q: i( v7 }+ {8 O4 w: Y3 k 
7 C3 f' N( p$ V, ` 
def Baidu_ocr(binary):% A1 \: v, N% J3 o 
    url = "https://aip.baidubce.com/oauth/2.0/token?" 
' j% {: @& J\" p: o' O; v, O    data = {* b0 g\" s2 Z) K. L; f8 } 
        "grant_type": "client_credentials",, c/ s\" t( V! x 
        "client_id": "",  # 必填2 i1 u8 o  ?& o3 l* l 
        "client_secret": ""  # 必填 
/ L& ^5 I6 {* c, \3 E9 K    } 
) `) G* a3 e. U4 D5 w( i; O    url += urlencode(data) 
: q* ?* g  n5 I& }* N. P; C7 [8 [    response = requests.get(url=url).json()( r( N* n! R, C 
    access_token = response['access_token'] 
! S; X\" H# B* ^ 
# P; I& ~) W0 y( b5 r    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token  U9 Q  ~1 u+ o3 ~3 y3 t 
    post_data = {- k: N% I9 t4 I( w. T\" d 
        "image": base64.b64encode(binary),# a2 E; `6 P8 ~# U- K' X 
        "url": "",8 i& c, ^- A1 Y) F5 d4 f+ T+ S) s, R 
        "language_type": "CHN_ENG", 
6 N1 K9 K\" n; b7 i8 o1 @' Y- G) r8 N        "detect_direction": "false", 
4 u1 [7 z\" x* L& _1 V6 A        "detect_language": "false",! q- t\" O$ u% d* f 
        "probability": "false"% q8 L2 A8 U2 c2 D; ` 
    }' m$ v& Q. q* ]* V: V; }* {9 [ 
    response = requests.post(url=url, data=post_data).json() 
1 g3 k# V2 {; m, T3 w  [! W( G\" p    num = response['words_result_num']: Z( R) T( k\" r$ Y 
    result = ''' ?1 u. ]8 R# }4 s9 N 
    for i in range(num):0 N; ~: w& n4 M- }! O. ^ 
        result += response['words_result'][ i]['words'] + "\r\n") k3 f5 F2 o! O, M 
    return result5 ^0 w) }2 F1 O) F! l2 u 
 
2 o3 Z\" V\" s, F\" \5 Z, s% O 
0 _5 ~, \- M. Y! X2 |; wif __name__ == "__main__":5 X3 J8 L  S2 Z. N\" g* Y 
    file_path = Dialog_box() 
- n- F3 N\" {8 z    if exists(file_path): 
5 N9 ^5 r) }/ J* I        data = Read_file(file_path) 
5 v. b- \# _2 h  L- \\" [  w        print(Baidu_ocr(data)). W\" b6 S, l$ l2 b2 Q; u( b, n9 H 
. g  i; A: b( @( ^7 j6 v

转发自派生社区

Python交流群：1047602540

zan