查看: 2897|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) V$ ~, O3 T. |\" A
# -*- coding: utf-8 -*-1 U! B9 R* g4 h7 B- F4 t
6 H( ]6 G8 j0 y
import base64
\" {5 A, W4 K3 Z! d- N' rfrom os.path import exists0 j, N/ i D: W
from tkinter import Tk
' {8 j L% d3 o- tfrom tkinter.filedialog import askopenfilename
! r# W1 D5 D& ^9 |/ }$ w* qfrom urllib.parse import urlencode3 `: {8 }2 A: P* }. p' Q) }1 t
( }2 f; d. j4 ~& ?8 `+ y\" S/ Z
import requests7 y3 e5 l; }% }0 E4 K! r8 L
@+ W) e& ^/ {1 A2 R/ @\" q- l$ h5 I0 v7 v
9 E1 g/ v; d. i/ a! l5 c) n+ adef Dialog_box():
\" H0 G& W8 J4 Q, S5 Q- p root = Tk()
. o' @, ]( m0 N! c root.withdraw()( V& N: n6 e! q; u, `- I2 @
9 q6 {/ @% \4 Y/ Q, _ filetypes = [(
! W9 S- |$ e( b- C, G4 v '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
\" T. H2 `! C/ I( O j ('所有文件(*.*)', '*.*')]2 z7 W T/ j. x* h8 Z$ ^
3 t6 P0 P\" D& h file_path = askopenfilename(filetypes=filetypes)
% C: z$ Z, r# J return file_path
1 e- H( f1 f2 {+ w- g2 s4 N* }
) c5 I$ w$ C$ d! h0 R' P
\" i. Z# x8 C; qdef Read_file(file_path):* D1 U) r5 ?* U! |8 h
with open(file_path, mode='rb') as file_object:
8 C0 V8 l* M6 y5 N contents = file_object.read()8 F; \9 I/ v S4 L6 f) U
return contents.rstrip()$ _7 o0 ~' u\" [3 u4 X, d
. u& }7 \0 W6 ]
R( \. T2 M+ o3 m4 [def Baidu_ocr(binary):% g7 k6 a1 t% Q5 w7 T\" y6 Z
url = "https://aip.baidubce.com/oauth/2.0/token?"
. K1 `7 M; Y4 @ data = {) H2 s6 f9 A5 W+ j) ^\" R6 F2 E
"grant_type": "client_credentials",\" n3 f) U8 r- m
"client_id": "", # 必填- ?, V E2 v, r j6 ^0 N
"client_secret": "" # 必填
# {* q+ q A) E& C }1 U: {9 g2 K- a8 c& \
url += urlencode(data)1 z2 F6 I: P. d( G
response = requests.get(url=url).json()
: {5 m7 W U- [! u, m access_token = response['access_token']
/ }, L; `* b9 ]' g2 E* T
; \\" }& n# J8 K+ b url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
5 z: C( p: y/ k& y {% g9 E/ A post_data = {
- t }4 P* O& Q! Z2 D' d "image": base64.b64encode(binary),
\" ]* K\" `- X, D& T; |\" ? "url": "",3 \& x0 A2 ^5 y4 w\" `
"language_type": "CHN_ENG",+ A# W$ y( y8 I7 O8 S4 S# L7 Z, W1 k% z
"detect_direction": "false",+ Z\" q5 P# q9 e/ C ^* U
"detect_language": "false",
8 G6 |5 i- s2 ?8 B5 D% I7 v( u "probability": "false"
& q% l) N- q# Z3 n }, J1 g9 }, P8 c) y) | m; F/ @
response = requests.post(url=url, data=post_data).json()
% r$ _* y4 ~: m! U num = response['words_result_num']
3 p4 m5 m+ x% L) m result = '') E' @; s6 j. c, ~
for i in range(num):, L0 M) B0 O! v/ m5 {8 M# j8 Z
result += response['words_result'][ i]['words'] + "\r\n"* T {7 r2 ^% \9 r3 Y9 P
return result, s; {\" T A) c
' F# z' _' }, s1 p! g( h
* ]& J4 z- b6 J2 ~5 ]6 F
if __name__ == "__main__":
: @0 [+ [% A* @/ x file_path = Dialog_box()% @+ Y- p6 E) K# ?! p; y* C' d8 t
if exists(file_path):, D/ E) h$ _# A; c f' u% }6 G, E
data = Read_file(file_path)2 M- M( Y$ S6 }2 d2 t8 }
print(Baidu_ocr(data))( B! H/ V. _\" e1 ]7 p, Z
8 n! I/ ^) x( O

#!/usr/bin/python) V$ ~, O3 T. |\" A 
# -*- coding: utf-8 -*-1 U! B9 R* g4 h7 B- F4 t 
6 H( ]6 G8 j0 y 
import base64 
\" {5 A, W4 K3 Z! d- N' rfrom os.path import exists0 j, N/ i  D: W 
from tkinter import Tk 
' {8 j  L% d3 o- tfrom tkinter.filedialog import askopenfilename 
! r# W1 D5 D& ^9 |/ }$ w* qfrom urllib.parse import urlencode3 `: {8 }2 A: P* }. p' Q) }1 t 
( }2 f; d. j4 ~& ?8 `+ y\" S/ Z 
import requests7 y3 e5 l; }% }0 E4 K! r8 L 
  @+ W) e& ^/ {1 A2 R/ @\" q- l$ h5 I0 v7 v 
 
9 E1 g/ v; d. i/ a! l5 c) n+ adef Dialog_box(): 
\" H0 G& W8 J4 Q, S5 Q- p    root = Tk() 
. o' @, ]( m0 N! c    root.withdraw()( V& N: n6 e! q; u, `- I2 @ 
 
9 q6 {/ @% \4 Y/ Q, _    filetypes = [( 
! W9 S- |$ e( b- C, G4 v        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
\" T. H2 `! C/ I( O  j        ('所有文件(*.*)', '*.*')]2 z7 W  T/ j. x* h8 Z$ ^ 
 
3 t6 P0 P\" D& h    file_path = askopenfilename(filetypes=filetypes) 
% C: z$ Z, r# J    return file_path 
1 e- H( f1 f2 {+ w- g2 s4 N* } 
) c5 I$ w$ C$ d! h0 R' P 
\" i. Z# x8 C; qdef Read_file(file_path):* D1 U) r5 ?* U! |8 h 
    with open(file_path, mode='rb') as file_object: 
8 C0 V8 l* M6 y5 N        contents = file_object.read()8 F; \9 I/ v  S4 L6 f) U 
        return contents.rstrip()$ _7 o0 ~' u\" [3 u4 X, d 
. u& }7 \0 W6 ] 
 
  R( \. T2 M+ o3 m4 [def Baidu_ocr(binary):% g7 k6 a1 t% Q5 w7 T\" y6 Z 
    url = "https://aip.baidubce.com/oauth/2.0/token?" 
. K1 `7 M; Y4 @    data = {) H2 s6 f9 A5 W+ j) ^\" R6 F2 E 
        "grant_type": "client_credentials",\" n3 f) U8 r- m 
        "client_id": "",  # 必填- ?, V  E2 v, r  j6 ^0 N 
        "client_secret": ""  # 必填 
# {* q+ q  A) E& C    }1 U: {9 g2 K- a8 c& \ 
    url += urlencode(data)1 z2 F6 I: P. d( G 
    response = requests.get(url=url).json() 
: {5 m7 W  U- [! u, m    access_token = response['access_token'] 
/ }, L; `* b9 ]' g2 E* T 
; \\" }& n# J8 K+ b    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
5 z: C( p: y/ k& y  {% g9 E/ A    post_data = { 
- t  }4 P* O& Q! Z2 D' d        "image": base64.b64encode(binary), 
\" ]* K\" `- X, D& T; |\" ?        "url": "",3 \& x0 A2 ^5 y4 w\" ` 
        "language_type": "CHN_ENG",+ A# W$ y( y8 I7 O8 S4 S# L7 Z, W1 k% z 
        "detect_direction": "false",+ Z\" q5 P# q9 e/ C  ^* U 
        "detect_language": "false", 
8 G6 |5 i- s2 ?8 B5 D% I7 v( u        "probability": "false" 
& q% l) N- q# Z3 n    }, J1 g9 }, P8 c) y) |  m; F/ @ 
    response = requests.post(url=url, data=post_data).json() 
% r$ _* y4 ~: m! U    num = response['words_result_num'] 
3 p4 m5 m+ x% L) m    result = '') E' @; s6 j. c, ~ 
    for i in range(num):, L0 M) B0 O! v/ m5 {8 M# j8 Z 
        result += response['words_result'][ i]['words'] + "\r\n"* T  {7 r2 ^% \9 r3 Y9 P 
    return result, s; {\" T  A) c 
' F# z' _' }, s1 p! g( h 
* ]& J4 z- b6 J2 ~5 ]6 F 
if __name__ == "__main__": 
: @0 [+ [% A* @/ x    file_path = Dialog_box()% @+ Y- p6 E) K# ?! p; y* C' d8 t 
    if exists(file_path):, D/ E) h$ _# A; c  f' u% }6 G, E 
        data = Read_file(file_path)2 M- M( Y$ S6 }2 d2 t8 } 
        print(Baidu_ocr(data))( B! H/ V. _\" e1 ]7 p, Z 
 
8 n! I/ ^) x( O

转发自派生社区

Python交流群：1047602540

zan