查看: 2839|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
3 B1 }1 ^6 \; ?# -*- coding: utf-8 -*-: Q( f _) v/ f6 ?/ c
3 m, i5 F( r) J Nimport base64
- s( o% _! }: E+ n/ u1 hfrom os.path import exists
, r& y& D( r\" c& u1 f1 f$ p+ Sfrom tkinter import Tk, A# }, v% \$ v! y$ \- J% q Z- F
from tkinter.filedialog import askopenfilename
- N$ J/ }9 W: X# _8 b! Z7 D; sfrom urllib.parse import urlencode
\" ?! ~7 ~7 G, b' Q* j5 b- i B: \+ o% p# {1 e
import requests
N8 ~\" n0 m& Z& I! ?! m3 G
W3 t ?' t6 S# y
. u; Z3 }) l) K- e9 @def Dialog_box():
7 s: n8 ~+ G9 X. O root = Tk()' t) `2 u$ V& \1 b& }9 s
root.withdraw()6 [* r5 V; j* s8 i% f5 S/ r
\" `1 g2 M0 o/ C3 X7 G, y( W5 e filetypes = [(+ K0 a# |; q7 N( ?# o/ I
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),, ^- i( o% r/ S/ Q6 g
('所有文件(*.*)', '*.*')]
' W/ L) E8 D3 U
9 @( h& e' \4 k$ y% o4 B file_path = askopenfilename(filetypes=filetypes)
2 m. A; x: N) P return file_path n$ {2 h( s5 Y; Z7 j5 J0 r
, R' f8 ~8 R9 T' r7 \1 O
- b9 W7 D! Q% Z\" n Hdef Read_file(file_path):5 Q% ^9 s6 }: E1 \ U
with open(file_path, mode='rb') as file_object:# T0 a( k9 T& y) `
contents = file_object.read(). L# j\" Y% x* a0 E z( v
return contents.rstrip()
, V, {+ a# }& _) I+ h4 ^
$ B7 v! k8 U* W& J# w
3 E: s1 Y B3 Hdef Baidu_ocr(binary):& @9 e, W. i/ ~7 d$ O\" w& c
url = "https://aip.baidubce.com/oauth/2.0/token?"( m\" R4 G1 P- z: U, S! b/ T$ P1 O% y+ y
data = {
+ |! Q# ^+ O7 h* i; L "grant_type": "client_credentials", N7 a. j& I- q3 v- I
"client_id": "", # 必填
5 g+ A2 e+ x2 ?2 T "client_secret": "" # 必填
. |, j. z0 s: _ }* L0 X' T+ I: C( A
url += urlencode(data)
% p\" G: N$ v; ~! s1 }$ m8 u response = requests.get(url=url).json()4 r% Q4 e3 g3 L5 N* i( s& d N
access_token = response['access_token']4 s\" {+ @! N' G& m3 y: V
\" D\" N& Y! Z; v% g i. b
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token# C: }) y( y+ \( M k& U! K
post_data = { X' T: y\" ?' J* r; H1 y
"image": base64.b64encode(binary),: R, Z, C* A) X- T& V
"url": "",
5 `\" r* s \4 F2 P+ n5 M "language_type": "CHN_ENG",) a: n- S: J! M, j! y' R
"detect_direction": "false",
; ~' H# y2 M' H2 R& r "detect_language": "false",2 d4 Z6 s! `1 b( I
"probability": "false"3 ~3 l1 f* [! O5 n% d! M! y
}
3 r2 h, h* r9 `5 N\" L' R3 M' E) m: O response = requests.post(url=url, data=post_data).json()) C9 g& Y. l! @
num = response['words_result_num']
4 y$ |* d3 @- K3 b' W1 m4 d9 n result = ''
1 g8 B1 Y7 g+ P6 ] for i in range(num):
|2 A: H5 l S8 \4 T result += response['words_result'][ i]['words'] + "\r\n"0 ?+ d- z/ q$ ]- n* H0 {+ \4 o
return result: W/ f. @6 b+ b G- i3 e4 \* j\" M2 p
1 k) y# J1 P9 Z& r6 u! l
! c: I1 X! F6 i' ]! [
if __name__ == "__main__":; L. I7 }3 @) X8 @3 o4 u( D
file_path = Dialog_box()* H& `- [5 X9 U1 z! H0 A2 q
if exists(file_path):: y* Z* w/ N% f& x9 L4 u
data = Read_file(file_path)9 W3 |: e& I% i& F }6 ?; w
print(Baidu_ocr(data))' @ r& f' Y\" ~2 \$ S& }& N
o: R. A& ]+ z0 }- P\" Z+ ` k

#!/usr/bin/python 
3 B1 }1 ^6 \; ?# -*- coding: utf-8 -*-: Q( f  _) v/ f6 ?/ c 
 
3 m, i5 F( r) J  Nimport base64 
- s( o% _! }: E+ n/ u1 hfrom os.path import exists 
, r& y& D( r\" c& u1 f1 f$ p+ Sfrom tkinter import Tk, A# }, v% \$ v! y$ \- J% q  Z- F 
from tkinter.filedialog import askopenfilename 
- N$ J/ }9 W: X# _8 b! Z7 D; sfrom urllib.parse import urlencode 
\" ?! ~7 ~7 G, b' Q* j5 b- i  B: \+ o% p# {1 e 
import requests 
  N8 ~\" n0 m& Z& I! ?! m3 G 
  W3 t  ?' t6 S# y 
. u; Z3 }) l) K- e9 @def Dialog_box(): 
7 s: n8 ~+ G9 X. O    root = Tk()' t) `2 u$ V& \1 b& }9 s 
    root.withdraw()6 [* r5 V; j* s8 i% f5 S/ r 
 
\" `1 g2 M0 o/ C3 X7 G, y( W5 e    filetypes = [(+ K0 a# |; q7 N( ?# o/ I 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),, ^- i( o% r/ S/ Q6 g 
        ('所有文件(*.*)', '*.*')] 
' W/ L) E8 D3 U 
9 @( h& e' \4 k$ y% o4 B    file_path = askopenfilename(filetypes=filetypes) 
2 m. A; x: N) P    return file_path  n$ {2 h( s5 Y; Z7 j5 J0 r 
 
, R' f8 ~8 R9 T' r7 \1 O 
- b9 W7 D! Q% Z\" n  Hdef Read_file(file_path):5 Q% ^9 s6 }: E1 \  U 
    with open(file_path, mode='rb') as file_object:# T0 a( k9 T& y) ` 
        contents = file_object.read(). L# j\" Y% x* a0 E  z( v 
        return contents.rstrip() 
, V, {+ a# }& _) I+ h4 ^ 
$ B7 v! k8 U* W& J# w 
3 E: s1 Y  B3 Hdef Baidu_ocr(binary):& @9 e, W. i/ ~7 d$ O\" w& c 
    url = "https://aip.baidubce.com/oauth/2.0/token?"( m\" R4 G1 P- z: U, S! b/ T$ P1 O% y+ y 
    data = { 
+ |! Q# ^+ O7 h* i; L        "grant_type": "client_credentials",  N7 a. j& I- q3 v- I 
        "client_id": "",  # 必填 
5 g+ A2 e+ x2 ?2 T        "client_secret": ""  # 必填 
. |, j. z0 s: _    }* L0 X' T+ I: C( A 
    url += urlencode(data) 
% p\" G: N$ v; ~! s1 }$ m8 u    response = requests.get(url=url).json()4 r% Q4 e3 g3 L5 N* i( s& d  N 
    access_token = response['access_token']4 s\" {+ @! N' G& m3 y: V 
\" D\" N& Y! Z; v% g  i. b 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token# C: }) y( y+ \( M  k& U! K 
    post_data = {  X' T: y\" ?' J* r; H1 y 
        "image": base64.b64encode(binary),: R, Z, C* A) X- T& V 
        "url": "", 
5 `\" r* s  \4 F2 P+ n5 M        "language_type": "CHN_ENG",) a: n- S: J! M, j! y' R 
        "detect_direction": "false", 
; ~' H# y2 M' H2 R& r        "detect_language": "false",2 d4 Z6 s! `1 b( I 
        "probability": "false"3 ~3 l1 f* [! O5 n% d! M! y 
    } 
3 r2 h, h* r9 `5 N\" L' R3 M' E) m: O    response = requests.post(url=url, data=post_data).json()) C9 g& Y. l! @ 
    num = response['words_result_num'] 
4 y$ |* d3 @- K3 b' W1 m4 d9 n    result = '' 
1 g8 B1 Y7 g+ P6 ]    for i in range(num): 
  |2 A: H5 l  S8 \4 T        result += response['words_result'][ i]['words'] + "\r\n"0 ?+ d- z/ q$ ]- n* H0 {+ \4 o 
    return result: W/ f. @6 b+ b  G- i3 e4 \* j\" M2 p 
1 k) y# J1 P9 Z& r6 u! l 
! c: I1 X! F6 i' ]! [ 
if __name__ == "__main__":; L. I7 }3 @) X8 @3 o4 u( D 
    file_path = Dialog_box()* H& `- [5 X9 U1 z! H0 A2 q 
    if exists(file_path):: y* Z* w/ N% f& x9 L4 u 
        data = Read_file(file_path)9 W3 |: e& I% i& F  }6 ?; w 
        print(Baidu_ocr(data))' @  r& f' Y\" ~2 \$ S& }& N 
  o: R. A& ]+ z0 }- P\" Z+ `  k

转发自派生社区

Python交流群：1047602540

zan