查看: 2776|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
- d: ?/ N7 ?, Q5 m# -*- coding: utf-8 -*-7 |9 s4 G0 i6 o6 k8 r \% T
8 `4 V6 d+ h; @* u& { [import base64' O; n* t5 s& O6 X
from os.path import exists% h; \ ?\" Y1 }, T& }) E/ J, g% |
from tkinter import Tk* O8 b% u q: G) D# ]! g. d
from tkinter.filedialog import askopenfilename0 S( y- { C1 \2 D$ k: N, W6 V
from urllib.parse import urlencode( a* |: }1 n3 I }, b- _# c
4 Q5 I6 Y' I* R
import requests
% f$ P' G4 T1 B$ y4 m( ]* C0 `: B
- c; b! l! C8 \9 z4 M8 C Q
! p6 t. l7 |& w' ~3 \def Dialog_box():# D; q' [6 t* q) X# U
root = Tk()
4 X8 ?: {1 Z& }+ M0 g( p& E$ J root.withdraw()1 T/ p! _0 x1 _. t8 I& S
/ b! \; M1 @, }8 l; s filetypes = [(; o\" E; U, H' j
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),' N) D/ Y) O3 T! l0 y. W; Y5 x
('所有文件(*.*)', '*.*')]7 X4 Z8 v; k7 O4 O
0 c9 e- Y: v7 I% u `3 L
file_path = askopenfilename(filetypes=filetypes)
2 L' F' R6 x$ z+ ^% Y return file_path
$ ?5 }1 z+ v/ Q
8 X$ g! O* I+ e\" e) |! ?5 z9 F6 ~1 E' u+ _/ N. P
def Read_file(file_path):
. r. u9 |+ @0 V with open(file_path, mode='rb') as file_object:\" n5 _! J9 Y5 I; N9 K6 T3 p2 |\" a
contents = file_object.read()
/ Z* @5 P6 U2 P0 @# D return contents.rstrip()
$ h b. A) ?, M) J7 @7 B8 F) `6 [* q( D) i, H; u+ C7 ~
: M* o$ n) s& l: L7 f. {
def Baidu_ocr(binary):+ l8 U `- v# s( a7 S
url = "https://aip.baidubce.com/oauth/2.0/token?"$ z( l$ ~3 x1 B* h' {5 Y& G8 {7 |/ z
data = {
3 k% l, v, R\" Q0 {% x1 H/ ] "grant_type": "client_credentials",
$ k/ E) b, m/ r2 ^7 A1 W "client_id": "", # 必填- M# ?+ Y1 g( j: d+ @# F4 S
"client_secret": "" # 必填
% Q9 A\" \0 w( u! \; L }) ]( e L; c q$ v\" u$ u; u
url += urlencode(data)
' g; g, @3 u# b$ k d, I* O2 Q) R response = requests.get(url=url).json() t* l; |7 u+ W( q6 K! D4 N( a8 o5 U
access_token = response['access_token']# t\" K1 Z- P# e: ]. j. ]
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
- t- V* Y+ l, A/ [* G* I) n: m# M post_data = {
& n! D: l5 L( [/ |9 P. G. T4 \ "image": base64.b64encode(binary),& f K6 e# w1 s\" q7 Z/ j
"url": "",
7 \\" k# ~) k5 I& `1 s, z) H+ S1 W "language_type": "CHN_ENG",* H M) k0 W) s/ w
"detect_direction": "false",
( w6 m. L( n3 m$ Y$ W+ W: D "detect_language": "false",
) \5 W) _6 F: a! F "probability": "false"
0 }* ^1 a0 y6 |4 W1 ~, f }
3 `4 N# J( k0 ]/ g& k$ V5 q\" q response = requests.post(url=url, data=post_data).json(); D; P, \4 k9 {* y9 _# u
num = response['words_result_num']\" H4 V& [% l9 n7 c1 D x
result = ''
4 n9 w t, x p( F for i in range(num):, c g1 Q0 A\" b. L) A& d
result += response['words_result'][ i]['words'] + "\r\n") [\" Z- x- o. }# p0 b1 }
return result$ i0 D6 x/ G r+ i
1 `\" N8 h1 N5 U\" L- d9 ]( K
) Y# h9 S9 ~' b5 Lif __name__ == "__main__":& ?/ Q) u( B\" y6 K) _3 j
file_path = Dialog_box(): H& d- @3 n- v0 n+ w) l
if exists(file_path):( o2 z D \2 z9 [
data = Read_file(file_path)
* z* d0 q8 m3 Y* ~: |, j$ v print(Baidu_ocr(data))
' M6 j; M$ D- d: i Y2 M\" c\" ]+ c( r4 ], x4 x- ?

#!/usr/bin/python 
- d: ?/ N7 ?, Q5 m# -*- coding: utf-8 -*-7 |9 s4 G0 i6 o6 k8 r  \% T 
 
8 `4 V6 d+ h; @* u& {  [import base64' O; n* t5 s& O6 X 
from os.path import exists% h; \  ?\" Y1 }, T& }) E/ J, g% | 
from tkinter import Tk* O8 b% u  q: G) D# ]! g. d 
from tkinter.filedialog import askopenfilename0 S( y- {  C1 \2 D$ k: N, W6 V 
from urllib.parse import urlencode( a* |: }1 n3 I  }, b- _# c 
4 Q5 I6 Y' I* R 
import requests 
% f$ P' G4 T1 B$ y4 m( ]* C0 `: B 
- c; b! l! C8 \9 z4 M8 C  Q 
! p6 t. l7 |& w' ~3 \def Dialog_box():# D; q' [6 t* q) X# U 
    root = Tk() 
4 X8 ?: {1 Z& }+ M0 g( p& E$ J    root.withdraw()1 T/ p! _0 x1 _. t8 I& S 
 
/ b! \; M1 @, }8 l; s    filetypes = [(; o\" E; U, H' j 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),' N) D/ Y) O3 T! l0 y. W; Y5 x 
        ('所有文件(*.*)', '*.*')]7 X4 Z8 v; k7 O4 O 
0 c9 e- Y: v7 I% u  `3 L 
    file_path = askopenfilename(filetypes=filetypes) 
2 L' F' R6 x$ z+ ^% Y    return file_path 
$ ?5 }1 z+ v/ Q 
8 X$ g! O* I+ e\" e) |! ?5 z9 F6 ~1 E' u+ _/ N. P 
def Read_file(file_path): 
. r. u9 |+ @0 V    with open(file_path, mode='rb') as file_object:\" n5 _! J9 Y5 I; N9 K6 T3 p2 |\" a 
        contents = file_object.read() 
/ Z* @5 P6 U2 P0 @# D        return contents.rstrip() 
$ h  b. A) ?, M) J7 @7 B8 F) `6 [* q( D) i, H; u+ C7 ~ 
: M* o$ n) s& l: L7 f. { 
def Baidu_ocr(binary):+ l8 U  `- v# s( a7 S 
    url = "https://aip.baidubce.com/oauth/2.0/token?"$ z( l$ ~3 x1 B* h' {5 Y& G8 {7 |/ z 
    data = { 
3 k% l, v, R\" Q0 {% x1 H/ ]        "grant_type": "client_credentials", 
$ k/ E) b, m/ r2 ^7 A1 W        "client_id": "",  # 必填- M# ?+ Y1 g( j: d+ @# F4 S 
        "client_secret": ""  # 必填 
% Q9 A\" \0 w( u! \; L    }) ]( e  L; c  q$ v\" u$ u; u 
    url += urlencode(data) 
' g; g, @3 u# b$ k  d, I* O2 Q) R    response = requests.get(url=url).json()  t* l; |7 u+ W( q6 K! D4 N( a8 o5 U 
    access_token = response['access_token']# t\" K1 Z- P# e: ]. j. ] 
 
1 V( b8 j0 s6 e1 U$ I. O  _: `- e$ v  y    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
- t- V* Y+ l, A/ [* G* I) n: m# M    post_data = { 
& n! D: l5 L( [/ |9 P. G. T4 \        "image": base64.b64encode(binary),& f  K6 e# w1 s\" q7 Z/ j 
        "url": "", 
7 \\" k# ~) k5 I& `1 s, z) H+ S1 W        "language_type": "CHN_ENG",* H  M) k0 W) s/ w 
        "detect_direction": "false", 
( w6 m. L( n3 m$ Y$ W+ W: D        "detect_language": "false", 
) \5 W) _6 F: a! F        "probability": "false" 
0 }* ^1 a0 y6 |4 W1 ~, f    } 
3 `4 N# J( k0 ]/ g& k$ V5 q\" q    response = requests.post(url=url, data=post_data).json(); D; P, \4 k9 {* y9 _# u 
    num = response['words_result_num']\" H4 V& [% l9 n7 c1 D  x 
    result = '' 
4 n9 w  t, x  p( F    for i in range(num):, c  g1 Q0 A\" b. L) A& d 
        result += response['words_result'][ i]['words'] + "\r\n") [\" Z- x- o. }# p0 b1 } 
    return result$ i0 D6 x/ G  r+ i 
 
1 `\" N8 h1 N5 U\" L- d9 ]( K 
) Y# h9 S9 ~' b5 Lif __name__ == "__main__":& ?/ Q) u( B\" y6 K) _3 j 
    file_path = Dialog_box(): H& d- @3 n- v0 n+ w) l 
    if exists(file_path):( o2 z  D  \2 z9 [ 
        data = Read_file(file_path) 
* z* d0 q8 m3 Y* ~: |, j$ v        print(Baidu_ocr(data)) 
' M6 j; M$ D- d: i  Y2 M\" c\" ]+ c( r4 ], x4 x- ?

转发自派生社区

Python交流群：1047602540

zan