查看: 2786|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
4 ?3 q) [& ?5 ~0 H, z5 I9 t# -*- coding: utf-8 -*-* L1 H/ [' o# M
# a$ k! K; I. rimport base64
4 h/ g5 X( `2 _7 ifrom os.path import exists
% o( J/ V6 h8 M* H: j B3 f# ^ ifrom tkinter import Tk: [0 c! R4 c$ b
from tkinter.filedialog import askopenfilename# Z8 c7 [, y3 o; e s\" a
from urllib.parse import urlencode' Z' q8 |; _1 m. }0 z8 O7 P
) u% D) w( ~. P6 J9 Himport requests
% |$ A8 W- x/ _# C# B
2 P4 w3 ]8 u6 M% N- o% g! V3 ~\" Q- P1 a: u! g. b. R) ]. ?( {5 R
def Dialog_box():
7 H0 _& E9 j$ x ]+ p3 Q$ h0 M root = Tk()
5 v1 g5 z7 L: Y2 s8 C6 L0 K9 z9 { root.withdraw()
- i' J/ p) k$ e2 f. J0 N# M) y\" x\" a% Q5 V m/ Z' W
filetypes = [(
2 g( e7 y. A S0 ?' k! H '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
2 T) ^. w4 Q. ?* L1 b% c5 ] ('所有文件(*.*)', '*.*')]: b( |; L2 X! D. d, @4 Q
8 @6 G l* E+ N X& z file_path = askopenfilename(filetypes=filetypes)
$ r. b, ^* Q! J: _ return file_path; d# G9 k8 I7 |0 O* O\" \
P8 N2 E% |& j- M k7 X: t1 x8 [: y7 ^- Q
def Read_file(file_path):/ F2 C. K( k0 m+ \; u7 ]% b3 S
with open(file_path, mode='rb') as file_object:( Y* r `6 l+ n+ s$ w# M* a, q
contents = file_object.read()
% C/ K3 N ^6 L9 s# P7 b return contents.rstrip()
% F8 ?9 | m4 D0 }- @+ i( h7 g; T6 D, f( s# ^
/ q& H2 E; \$ u2 L# Qdef Baidu_ocr(binary):
- l4 l+ N5 F; [5 S7 z0 h* r6 P url = "https://aip.baidubce.com/oauth/2.0/token?", B3 o# h\" {5 t. G! \
data = {
( \& e: c+ H6 b' A& L5 c8 C `& G "grant_type": "client_credentials",
; n2 n! C2 W/ H4 k* c! D "client_id": "", # 必填' O% g5 `) r. k7 `7 M
"client_secret": "" # 必填
* S) v1 M\" L' _5 K! k }; C: L3 [$ n& \4 ~3 T
url += urlencode(data)+ `* m/ X- ]( c\" U
response = requests.get(url=url).json()
( B8 O! {3 s) H% U$ b9 B! _ access_token = response['access_token']
. Q+ f# z, a! I; P0 v1 k
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token$ X6 \8 l0 ^' j% Z# {& ]8 a# t
post_data = {
! _! |5 L2 j# e "image": base64.b64encode(binary),) k' V( k% i5 W+ ^7 {2 u
"url": "",, q. ]* b& N$ l6 N5 w' }, S
"language_type": "CHN_ENG", Q\" f( V! O5 U* g! y$ k2 }% \
"detect_direction": "false",/ |8 U$ L3 f! Q, S
"detect_language": "false",
. c( U) l c$ C6 s "probability": "false"
/ z5 h5 @) I# _& H\" D9 T1 c }/ i4 f( h4 \8 k6 Q& ^( K2 C
response = requests.post(url=url, data=post_data).json()
\" |2 a+ `7 D4 y3 q& ]1 c% y num = response['words_result_num']
4 Q. s9 E9 w r$ j+ E. A result = ''
\" Q3 W7 \3 y5 a6 a5 [ for i in range(num):
\" V2 j! N1 s1 `1 J# i result += response['words_result'][ i]['words'] + "\r\n"# |6 j) k. }5 v2 A0 m/ o: i6 w$ P# H
return result
- q: X- ~- U% G2 j5 V
+ x, b, Z3 V2 \- t a* d9 b+ K; ?4 ]) m5 M9 W% g
if __name__ == "__main__":
4 n* G: j4 F' t! Y3 i/ {: { file_path = Dialog_box()
! n4 b) L4 s+ J- d$ j2 u' x1 U if exists(file_path):0 ]) [% c/ t1 A: ^7 f' i
data = Read_file(file_path). p3 _3 X+ N0 [3 z
print(Baidu_ocr(data))4 d6 s\" p& T; b* T P! t7 G
/ P6 a. ?6 u( y6 A$ V

#!/usr/bin/python 
4 ?3 q) [& ?5 ~0 H, z5 I9 t# -*- coding: utf-8 -*-* L1 H/ [' o# M 
 
# a$ k! K; I. rimport base64 
4 h/ g5 X( `2 _7 ifrom os.path import exists 
% o( J/ V6 h8 M* H: j  B3 f# ^  ifrom tkinter import Tk: [0 c! R4 c$ b 
from tkinter.filedialog import askopenfilename# Z8 c7 [, y3 o; e  s\" a 
from urllib.parse import urlencode' Z' q8 |; _1 m. }0 z8 O7 P 
 
) u% D) w( ~. P6 J9 Himport requests 
% |$ A8 W- x/ _# C# B 
2 P4 w3 ]8 u6 M% N- o% g! V3 ~\" Q- P1 a: u! g. b. R) ]. ?( {5 R 
def Dialog_box(): 
7 H0 _& E9 j$ x  ]+ p3 Q$ h0 M    root = Tk() 
5 v1 g5 z7 L: Y2 s8 C6 L0 K9 z9 {    root.withdraw() 
- i' J/ p) k$ e2 f. J0 N# M) y\" x\" a% Q5 V  m/ Z' W 
    filetypes = [( 
2 g( e7 y. A  S0 ?' k! H        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
2 T) ^. w4 Q. ?* L1 b% c5 ]        ('所有文件(*.*)', '*.*')]: b( |; L2 X! D. d, @4 Q 
 
8 @6 G  l* E+ N  X& z    file_path = askopenfilename(filetypes=filetypes) 
$ r. b, ^* Q! J: _    return file_path; d# G9 k8 I7 |0 O* O\" \ 
 
  P8 N2 E% |& j- M  k7 X: t1 x8 [: y7 ^- Q 
def Read_file(file_path):/ F2 C. K( k0 m+ \; u7 ]% b3 S 
    with open(file_path, mode='rb') as file_object:( Y* r  `6 l+ n+ s$ w# M* a, q 
        contents = file_object.read() 
% C/ K3 N  ^6 L9 s# P7 b        return contents.rstrip() 
% F8 ?9 |  m4 D0 }- @+ i( h7 g; T6 D, f( s# ^ 
 
/ q& H2 E; \$ u2 L# Qdef Baidu_ocr(binary): 
- l4 l+ N5 F; [5 S7 z0 h* r6 P    url = "https://aip.baidubce.com/oauth/2.0/token?", B3 o# h\" {5 t. G! \ 
    data = { 
( \& e: c+ H6 b' A& L5 c8 C  `& G        "grant_type": "client_credentials", 
; n2 n! C2 W/ H4 k* c! D        "client_id": "",  # 必填' O% g5 `) r. k7 `7 M 
        "client_secret": ""  # 必填 
* S) v1 M\" L' _5 K! k    }; C: L3 [$ n& \4 ~3 T 
    url += urlencode(data)+ `* m/ X- ]( c\" U 
    response = requests.get(url=url).json() 
( B8 O! {3 s) H% U$ b9 B! _    access_token = response['access_token'] 
. Q+ f# z, a! I; P0 v1 k 
% @% k/ j9 N1 R\" p\" G/ h' p4 v; k7 f* p    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token$ X6 \8 l0 ^' j% Z# {& ]8 a# t 
    post_data = { 
! _! |5 L2 j# e        "image": base64.b64encode(binary),) k' V( k% i5 W+ ^7 {2 u 
        "url": "",, q. ]* b& N$ l6 N5 w' }, S 
        "language_type": "CHN_ENG",  Q\" f( V! O5 U* g! y$ k2 }% \ 
        "detect_direction": "false",/ |8 U$ L3 f! Q, S 
        "detect_language": "false", 
. c( U) l  c$ C6 s        "probability": "false" 
/ z5 h5 @) I# _& H\" D9 T1 c    }/ i4 f( h4 \8 k6 Q& ^( K2 C 
    response = requests.post(url=url, data=post_data).json() 
\" |2 a+ `7 D4 y3 q& ]1 c% y    num = response['words_result_num'] 
4 Q. s9 E9 w  r$ j+ E. A    result = '' 
\" Q3 W7 \3 y5 a6 a5 [    for i in range(num): 
\" V2 j! N1 s1 `1 J# i        result += response['words_result'][ i]['words'] + "\r\n"# |6 j) k. }5 v2 A0 m/ o: i6 w$ P# H 
    return result 
- q: X- ~- U% G2 j5 V 
+ x, b, Z3 V2 \- t  a* d9 b+ K; ?4 ]) m5 M9 W% g 
if __name__ == "__main__": 
4 n* G: j4 F' t! Y3 i/ {: {    file_path = Dialog_box() 
! n4 b) L4 s+ J- d$ j2 u' x1 U    if exists(file_path):0 ]) [% c/ t1 A: ^7 f' i 
        data = Read_file(file_path). p3 _3 X+ N0 [3 z 
        print(Baidu_ocr(data))4 d6 s\" p& T; b* T  P! t7 G 
/ P6 a. ?6 u( y6 A$ V

转发自派生社区

Python交流群：1047602540

zan