查看: 2789|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: U6 L9 R' i' i
# -*- coding: utf-8 -*-4 X! x* ^% j6 N) r8 x
( V4 ]: p6 a/ l\" D2 `+ t9 V8 }9 Kimport base64
0 w6 I5 R0 ?$ ~3 }from os.path import exists
3 k% _+ x- u4 A5 _1 _% n3 kfrom tkinter import Tk1 \$ V) A$ t9 s( p( ]. u6 w6 ^# A! D
from tkinter.filedialog import askopenfilename# ]4 H C# o1 h9 V
from urllib.parse import urlencode% @& D* }9 c/ [( x4 W
* @- |+ j( }! `* C+ w
import requests: r/ m\" r5 L$ ]7 B8 l
6 r' _- i4 m! m! Q# h) ?) P# N. Z* g) m9 ?7 p* f
def Dialog_box():6 H/ u& r& |( U1 r$ s
root = Tk()! q% y8 U\" l' V% T; n
root.withdraw()1 |' S3 }) R% X
& G: d+ p* W3 C* X5 X5 S: u3 V
filetypes = [(- q: y. y& O( G7 j( c+ s
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),1 z- ^3 }! k+ _$ ]# B
('所有文件(*.*)', '*.*')]
) H, [- x S8 R' t% u+ ^& V- f8 W
file_path = askopenfilename(filetypes=filetypes)' \3 R9 _6 @\" ]6 l/ q( W' T
return file_path
' c. F2 x+ i) T) H k: p: a, m/ O0 S. {9 G, ~1 s) ~8 V\" G
; R7 F; }1 [2 h8 H# M. b$ sdef Read_file(file_path):+ ]% I; \* [3 I, W) J
with open(file_path, mode='rb') as file_object:
; Z; z0 z+ y- ~2 a' t3 N contents = file_object.read()8 q* ^ }# V0 x
return contents.rstrip()
8 _+ J, z9 h% Q) ]
) l0 S8 ?\" E1 l2 W a0 @3 A P# c) C, [: P: Z; G1 b9 [2 s, g3 W
def Baidu_ocr(binary):
: v6 [3 I! U7 k7 x+ s url = "https://aip.baidubce.com/oauth/2.0/token?"# K8 }* C; c+ w' ~' D) k( |3 b
data = {2 A, w: x\" \( i# ]. g3 ~
"grant_type": "client_credentials",5 A) e m1 I; @: B
"client_id": "", # 必填 O! ?; r' m p' Y1 a6 c, F
"client_secret": "" # 必填
! z$ v5 S# c+ w3 _ C+ R }
$ B9 l. `( A, _, t url += urlencode(data)1 n2 H' N# E. i0 y
response = requests.get(url=url).json()
0 }- |/ I6 R' X/ K0 K access_token = response['access_token']
& |9 f$ U9 p% `9 M G* r\" U* t0 u; R% A3 ~% ]: _- V
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token+ x* L; |5 ^( I& p# x
post_data = {; O, z( o5 ~( {
"image": base64.b64encode(binary),2 R0 E5 o, i4 {( c* M% |. J
"url": "",
/ ^7 a- G. l. i "language_type": "CHN_ENG",4 l3 p\" o; }9 |( Q
"detect_direction": "false",7 ^% B# b2 P3 Z. G
"detect_language": "false",9 t4 h \& f, G- \; O# y/ r0 t( N
"probability": "false"6 V6 {. b$ L9 E
}, B' w1 ]. X# _\" |7 z6 `' [
response = requests.post(url=url, data=post_data).json()
3 ^$ H; Q0 I: u/ B. C& `2 b num = response['words_result_num']
1 |9 k- x\" c w5 a8 L result = ''
+ E) [, p' j, m w# q' J3 e- `& M for i in range(num):
; @2 j2 k4 c9 X- n: Z9 o5 B5 H result += response['words_result'][ i]['words'] + "\r\n"2 Q9 }; _$ @, c( q# {$ u
return result
9 S+ a2 G, y! H) P* s3 \- T' m
5 g/ I, H1 P5 N e# e$ h7 T5 A1 O e- Q: G' a+ X
if __name__ == "__main__":; O5 s/ A- U6 b2 X8 v. @6 E
file_path = Dialog_box()5 s u& I3 Y; A; l
if exists(file_path):
+ u2 Y4 u& _4 g0 x5 X data = Read_file(file_path)( _* M8 T\" ?\" W$ g( Y, N
print(Baidu_ocr(data))
^) }1 Z* r3 x) D1 g2 u
& K, ? c6 Z5 Z! A

#!/usr/bin/python: U6 L9 R' i' i 
# -*- coding: utf-8 -*-4 X! x* ^% j6 N) r8 x 
 
( V4 ]: p6 a/ l\" D2 `+ t9 V8 }9 Kimport base64 
0 w6 I5 R0 ?$ ~3 }from os.path import exists 
3 k% _+ x- u4 A5 _1 _% n3 kfrom tkinter import Tk1 \$ V) A$ t9 s( p( ]. u6 w6 ^# A! D 
from tkinter.filedialog import askopenfilename# ]4 H  C# o1 h9 V 
from urllib.parse import urlencode% @& D* }9 c/ [( x4 W 
* @- |+ j( }! `* C+ w 
import requests: r/ m\" r5 L$ ]7 B8 l 
 
6 r' _- i4 m! m! Q# h) ?) P# N. Z* g) m9 ?7 p* f 
def Dialog_box():6 H/ u& r& |( U1 r$ s 
    root = Tk()! q% y8 U\" l' V% T; n 
    root.withdraw()1 |' S3 }) R% X 
& G: d+ p* W3 C* X5 X5 S: u3 V 
    filetypes = [(- q: y. y& O( G7 j( c+ s 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),1 z- ^3 }! k+ _$ ]# B 
        ('所有文件(*.*)', '*.*')] 
) H, [- x  S8 R' t% u+ ^& V- f8 W 
    file_path = askopenfilename(filetypes=filetypes)' \3 R9 _6 @\" ]6 l/ q( W' T 
    return file_path 
' c. F2 x+ i) T) H  k: p: a, m/ O0 S. {9 G, ~1 s) ~8 V\" G 
 
; R7 F; }1 [2 h8 H# M. b$ sdef Read_file(file_path):+ ]% I; \* [3 I, W) J 
    with open(file_path, mode='rb') as file_object: 
; Z; z0 z+ y- ~2 a' t3 N        contents = file_object.read()8 q* ^  }# V0 x 
        return contents.rstrip() 
8 _+ J, z9 h% Q) ] 
) l0 S8 ?\" E1 l2 W  a0 @3 A  P# c) C, [: P: Z; G1 b9 [2 s, g3 W 
def Baidu_ocr(binary): 
: v6 [3 I! U7 k7 x+ s    url = "https://aip.baidubce.com/oauth/2.0/token?"# K8 }* C; c+ w' ~' D) k( |3 b 
    data = {2 A, w: x\" \( i# ]. g3 ~ 
        "grant_type": "client_credentials",5 A) e  m1 I; @: B 
        "client_id": "",  # 必填  O! ?; r' m  p' Y1 a6 c, F 
        "client_secret": ""  # 必填 
! z$ v5 S# c+ w3 _  C+ R    } 
$ B9 l. `( A, _, t    url += urlencode(data)1 n2 H' N# E. i0 y 
    response = requests.get(url=url).json() 
0 }- |/ I6 R' X/ K0 K    access_token = response['access_token'] 
& |9 f$ U9 p% `9 M  G* r\" U* t0 u; R% A3 ~% ]: _- V 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token+ x* L; |5 ^( I& p# x 
    post_data = {; O, z( o5 ~( { 
        "image": base64.b64encode(binary),2 R0 E5 o, i4 {( c* M% |. J 
        "url": "", 
/ ^7 a- G. l. i        "language_type": "CHN_ENG",4 l3 p\" o; }9 |( Q 
        "detect_direction": "false",7 ^% B# b2 P3 Z. G 
        "detect_language": "false",9 t4 h  \& f, G- \; O# y/ r0 t( N 
        "probability": "false"6 V6 {. b$ L9 E 
    }, B' w1 ]. X# _\" |7 z6 `' [ 
    response = requests.post(url=url, data=post_data).json() 
3 ^$ H; Q0 I: u/ B. C& `2 b    num = response['words_result_num'] 
1 |9 k- x\" c  w5 a8 L    result = '' 
+ E) [, p' j, m  w# q' J3 e- `& M    for i in range(num): 
; @2 j2 k4 c9 X- n: Z9 o5 B5 H        result += response['words_result'][ i]['words'] + "\r\n"2 Q9 }; _$ @, c( q# {$ u 
    return result 
9 S+ a2 G, y! H) P* s3 \- T' m 
5 g/ I, H1 P5 N  e# e$ h7 T5 A1 O  e- Q: G' a+ X 
if __name__ == "__main__":; O5 s/ A- U6 b2 X8 v. @6 E 
    file_path = Dialog_box()5 s  u& I3 Y; A; l 
    if exists(file_path): 
+ u2 Y4 u& _4 g0 x5 X        data = Read_file(file_path)( _* M8 T\" ?\" W$ g( Y, N 
        print(Baidu_ocr(data)) 
  ^) }1 Z* r3 x) D1 g2 u 
& K, ?  c6 Z5 Z! A

转发自派生社区

Python交流群：1047602540

zan