查看: 2771|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
' ^, Q% ]5 B# U/ M& V8 X# -*- coding: utf-8 -*-' ?) |' |4 L3 x) Z0 i\" ]7 H/ A: R
/ t: b4 G( N! h* n: R' L% fimport base64
! e* M\" I: ^3 V/ }8 efrom os.path import exists6 Q/ x1 s/ a/ N% S
from tkinter import Tk
7 x, w+ w4 u( }0 J6 u Kfrom tkinter.filedialog import askopenfilename4 n9 Y6 R. M* x* k r+ g9 D0 h
from urllib.parse import urlencode# k, k$ r) r4 M4 U' s; V3 K
% C& r& f6 q/ I8 q fimport requests
& r# M8 \) p* ~5 I% x+ I- o3 u3 @, j9 P' X7 n
# z% J4 z+ M- O8 u4 p7 o: Jdef Dialog_box():/ B' ]$ d3 _( v7 q6 J9 b
root = Tk()
root.withdraw()/ k; F* c R: h0 F5 F, @ E! q
, x6 G4 F0 W. e' `; Y
filetypes = [(
5 X5 k+ Q- t+ z6 j: e '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
9 n# _\" N9 c\" z9 x3 m ('所有文件(*.*)', '*.*')]3 e# i: q7 p I
6 X3 K7 j' v- @0 I( B8 h8 z3 o( @ file_path = askopenfilename(filetypes=filetypes)
6 N7 F( j+ T) l return file_path
- S6 ?( P# ?4 O/ [5 h- y3 h8 h0 w% z, u. \ o# L) Y( b) T0 e! O
0 Q\" s3 k; T/ s7 @% Y1 h
def Read_file(file_path):
. a8 {3 f& f# U2 J with open(file_path, mode='rb') as file_object:
, o$ C0 d6 C9 K contents = file_object.read()2 z( `: S7 |2 A1 L+ n0 ~% i) k
return contents.rstrip()
% N7 Z0 h& {$ p( \( Y
7 H& E/ k( O8 l: k8 A8 c a: z0 D
def Baidu_ocr(binary):
- }8 x% ~: O3 F) S) g url = "https://aip.baidubce.com/oauth/2.0/token?"$ Y0 m6 s' V+ i- R
data = {- x: e& ]8 m3 C\" ]4 S% @
"grant_type": "client_credentials",
. x' W/ `- t/ F( u "client_id": "", # 必填
; n6 l' w( `% |) r5 P7 q "client_secret": "" # 必填
* B& ~* Y2 u% ]* j }
0 d) C- y$ Y/ l; `+ d url += urlencode(data)
7 Y/ _ }( J. K7 M response = requests.get(url=url).json()5 C& u& t$ o( M1 r
access_token = response['access_token'], r* `8 k; U( i
0 @9 r# S, I8 f: E8 B3 b url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token* P$ R' O9 f, J3 D
post_data = {8 k' V0 N- ]/ E
"image": base64.b64encode(binary),! h. D1 f% w) N
"url": "",
# K1 U4 m( e @ I9 x5 g "language_type": "CHN_ENG",, h5 ]8 E( \/ ?
"detect_direction": "false",
9 q8 P. R% t' x) D4 \ "detect_language": "false",
9 N3 F! H6 N2 M. {$ [/ B "probability": "false"
1 Q2 e$ `4 M0 p$ h8 O }9 x1 L7 U8 a. G! m% C. \& s% ]
response = requests.post(url=url, data=post_data).json()- h5 S8 O. k8 T$ M6 c! Y
num = response['words_result_num']# v' T+ i9 d! G' D g
result = ''
\" s* b& M9 e O4 y) c- y for i in range(num):
6 ^\" g- m\" q/ Z result += response['words_result'][ i]['words'] + "\r\n"
5 N! t- ~5 X\" T* v, T return result0 g9 O- R; f! \8 V4 ?, U
. B) u% ?$ b2 m; M* [2 A
$ k% V0 o# H* q( @& h/ B$ C/ kif __name__ == "__main__":
5 J* ^5 L7 ?2 W, [ file_path = Dialog_box()
\" m2 c/ _# z' s, u- h! L if exists(file_path):
; o( ]& B\" [4 Z5 Z- y, E data = Read_file(file_path)/ r4 x7 D8 x\" b+ ~9 q+ Z2 j
print(Baidu_ocr(data)): I2 ~, o9 \5 F( N2 l0 k
+ L7 E% u8 N; V% ~

#!/usr/bin/python 
' ^, Q% ]5 B# U/ M& V8 X# -*- coding: utf-8 -*-' ?) |' |4 L3 x) Z0 i\" ]7 H/ A: R 
 
/ t: b4 G( N! h* n: R' L% fimport base64 
! e* M\" I: ^3 V/ }8 efrom os.path import exists6 Q/ x1 s/ a/ N% S 
from tkinter import Tk 
7 x, w+ w4 u( }0 J6 u  Kfrom tkinter.filedialog import askopenfilename4 n9 Y6 R. M* x* k  r+ g9 D0 h 
from urllib.parse import urlencode# k, k$ r) r4 M4 U' s; V3 K 
 
% C& r& f6 q/ I8 q  fimport requests 
& r# M8 \) p* ~5 I% x+ I- o3 u3 @, j9 P' X7 n 
 
# z% J4 z+ M- O8 u4 p7 o: Jdef Dialog_box():/ B' ]$ d3 _( v7 q6 J9 b 
    root = Tk() 
. A* p% K$ S: e\" J# X4 M- f* p# H1 \* _    root.withdraw()/ k; F* c  R: h0 F5 F, @  E! q 
, x6 G4 F0 W. e' `; Y 
    filetypes = [( 
5 X5 k+ Q- t+ z6 j: e        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
9 n# _\" N9 c\" z9 x3 m        ('所有文件(*.*)', '*.*')]3 e# i: q7 p  I 
 
6 X3 K7 j' v- @0 I( B8 h8 z3 o( @    file_path = askopenfilename(filetypes=filetypes) 
6 N7 F( j+ T) l    return file_path 
- S6 ?( P# ?4 O/ [5 h- y3 h8 h0 w% z, u. \  o# L) Y( b) T0 e! O 
0 Q\" s3 k; T/ s7 @% Y1 h 
def Read_file(file_path): 
. a8 {3 f& f# U2 J    with open(file_path, mode='rb') as file_object: 
, o$ C0 d6 C9 K        contents = file_object.read()2 z( `: S7 |2 A1 L+ n0 ~% i) k 
        return contents.rstrip() 
% N7 Z0 h& {$ p( \( Y 
7 H& E/ k( O8 l: k8 A8 c  a: z0 D 
def Baidu_ocr(binary): 
- }8 x% ~: O3 F) S) g    url = "https://aip.baidubce.com/oauth/2.0/token?"$ Y0 m6 s' V+ i- R 
    data = {- x: e& ]8 m3 C\" ]4 S% @ 
        "grant_type": "client_credentials", 
. x' W/ `- t/ F( u        "client_id": "",  # 必填 
; n6 l' w( `% |) r5 P7 q        "client_secret": ""  # 必填 
* B& ~* Y2 u% ]* j    } 
0 d) C- y$ Y/ l; `+ d    url += urlencode(data) 
7 Y/ _  }( J. K7 M    response = requests.get(url=url).json()5 C& u& t$ o( M1 r 
    access_token = response['access_token'], r* `8 k; U( i 
 
0 @9 r# S, I8 f: E8 B3 b    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token* P$ R' O9 f, J3 D 
    post_data = {8 k' V0 N- ]/ E 
        "image": base64.b64encode(binary),! h. D1 f% w) N 
        "url": "", 
# K1 U4 m( e  @  I9 x5 g        "language_type": "CHN_ENG",, h5 ]8 E( \/ ? 
        "detect_direction": "false", 
9 q8 P. R% t' x) D4 \        "detect_language": "false", 
9 N3 F! H6 N2 M. {$ [/ B        "probability": "false" 
1 Q2 e$ `4 M0 p$ h8 O    }9 x1 L7 U8 a. G! m% C. \& s% ] 
    response = requests.post(url=url, data=post_data).json()- h5 S8 O. k8 T$ M6 c! Y 
    num = response['words_result_num']# v' T+ i9 d! G' D  g 
    result = '' 
\" s* b& M9 e  O4 y) c- y    for i in range(num): 
6 ^\" g- m\" q/ Z        result += response['words_result'][ i]['words'] + "\r\n" 
5 N! t- ~5 X\" T* v, T    return result0 g9 O- R; f! \8 V4 ?, U 
 
. B) u% ?$ b2 m; M* [2 A 
$ k% V0 o# H* q( @& h/ B$ C/ kif __name__ == "__main__": 
5 J* ^5 L7 ?2 W, [    file_path = Dialog_box() 
\" m2 c/ _# z' s, u- h! L    if exists(file_path): 
; o( ]& B\" [4 Z5 Z- y, E        data = Read_file(file_path)/ r4 x7 D8 x\" b+ ~9 q+ Z2 j 
        print(Baidu_ocr(data)): I2 ~, o9 \5 F( N2 l0 k 
+ L7 E% u8 N; V% ~

转发自派生社区

Python交流群：1047602540

zan