查看: 2813|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( e! K/ ?1 A _: `$ N
# -*- coding: utf-8 -*-
0 d% F8 u3 p+ c* U8 w# {9 m* Y L) }\" T4 C# @5 A+ i8 f) \0 d
import base642 c7 f) y- V\" y5 {0 U! `8 L
from os.path import exists
3 W: S' b1 d0 u5 ]7 F% Qfrom tkinter import Tk+ J& }* F' @1 _4 |
from tkinter.filedialog import askopenfilename
+ l\" @3 d5 y3 F& r4 r3 g+ efrom urllib.parse import urlencode5 e% |9 e9 p# N9 U: Y+ j8 [5 s
\" e) ` j _7 h3 y8 R3 Himport requests
0 ~5 h* i; S/ p5 s! d6 P4 y5 g+ G6 G7 B: h/ [! ?' J
; l5 ?' X' e5 C( c\" z& ~
def Dialog_box():6 _9 Q: D5 H2 `8 N4 u
root = Tk()/ s1 q3 S% g- g+ E) w
root.withdraw(). W* R8 u/ I& V! D2 f7 b
+ r+ `1 t- J d7 i( Z\" N2 `0 f: f
filetypes = [(3 j& E$ z2 ?7 |) J+ y/ y* z& b' m
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),; R2 I: {. R* {! N1 S x! G) Y7 |
('所有文件(*.*)', '*.*')]
& l4 f5 P+ j' p; e' P6 _) ]2 _1 o s8 K) n. I
file_path = askopenfilename(filetypes=filetypes)
6 H2 X/ Z- Z& @ return file_path# ~; T) _% v1 i8 X3 n
+ o; @! ~' m; P: H6 N5 f- i2 P6 N; w' N( `/ {# c5 ~4 ]
def Read_file(file_path):2 g1 I0 l' c( r
with open(file_path, mode='rb') as file_object:
5 ?/ T: t% I$ R4 j ]5 Q contents = file_object.read()
0 ?) C9 ]; b: B# [3 j( `$ d return contents.rstrip()* W9 Y; j* J& C3 d B% p
, p9 d' A' C\" H* L+ K6 m8 V
7 @4 g2 B' z+ |/ m- g9 N
def Baidu_ocr(binary):
/ o3 q8 \+ f- m6 | url = "https://aip.baidubce.com/oauth/2.0/token?"
- M! \3 @$ N: r- @ data = {. g. a& }+ B% x
"grant_type": "client_credentials",0 e8 v& K: F0 H! G\" ]: N7 K! T% \8 ?
"client_id": "", # 必填8 O7 Z2 v& _4 T, r+ c) d3 Q
"client_secret": "" # 必填5 d# r$ E: u9 @$ V; K3 [/ @1 d
}
/ E5 S* t; Y# E! `0 e5 ?4 L url += urlencode(data)
6 x0 {: n+ s. u2 C response = requests.get(url=url).json(); R4 N( g. A2 j! F4 [% X2 M
access_token = response['access_token']- k K- M8 T; T
7 S+ ^1 d6 m7 s- \9 S; u0 W5 Z url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
: K2 W$ A$ D/ t9 j$ L+ G post_data = {* V! k' x# }. G
"image": base64.b64encode(binary),2 H) q1 w/ ^: z9 G2 |' A+ Z
"url": "",
% M) l- o8 E+ E "language_type": "CHN_ENG",; J1 C, N h2 [& p7 `
"detect_direction": "false",5 O) E\" R X! u$ S
"detect_language": "false",) I: n. r3 d4 \+ y( O! K\" F! M
"probability": "false"% ?1 ?, i5 w ^) y
}
* O4 t+ b( z; g# m4 V response = requests.post(url=url, data=post_data).json()
0 X/ I/ g* I' t, @ p- Z num = response['words_result_num']
, X% J1 H' l' n- _( k$ t, x result = ''; l& _7 X- W/ ]) \
for i in range(num):
, v: }* Z8 E) |# j6 H) w6 p result += response['words_result'][ i]['words'] + "\r\n"
( @- V: n3 v$ Y\" | return result& I$ w7 s5 b- B
- J, I$ F5 z& f3 [8 j. K) p
\" N9 W( Y9 k\" y& X/ H% v1 I# ~5 j) eif __name__ == "__main__":
( _& t1 _2 b4 {3 _, w6 z file_path = Dialog_box()
\" h4 l# R s$ W, g- o, \$ R7 w& B if exists(file_path):
; A% y; X6 R- O/ K: ?. n data = Read_file(file_path)# P4 @\" H/ O' r! u, z. `7 V8 \! T
print(Baidu_ocr(data))\" G) w. J! U& \5 a1 `1 X( j
: s, K$ p2 k' d( T. R- M5 k) F9 G

#!/usr/bin/python( e! K/ ?1 A  _: `$ N 
# -*- coding: utf-8 -*- 
0 d% F8 u3 p+ c* U8 w# {9 m* Y  L) }\" T4 C# @5 A+ i8 f) \0 d 
import base642 c7 f) y- V\" y5 {0 U! `8 L 
from os.path import exists 
3 W: S' b1 d0 u5 ]7 F% Qfrom tkinter import Tk+ J& }* F' @1 _4 | 
from tkinter.filedialog import askopenfilename 
+ l\" @3 d5 y3 F& r4 r3 g+ efrom urllib.parse import urlencode5 e% |9 e9 p# N9 U: Y+ j8 [5 s 
 
\" e) `  j  _7 h3 y8 R3 Himport requests 
0 ~5 h* i; S/ p5 s! d6 P4 y5 g+ G6 G7 B: h/ [! ?' J 
; l5 ?' X' e5 C( c\" z& ~ 
def Dialog_box():6 _9 Q: D5 H2 `8 N4 u 
    root = Tk()/ s1 q3 S% g- g+ E) w 
    root.withdraw(). W* R8 u/ I& V! D2 f7 b 
+ r+ `1 t- J  d7 i( Z\" N2 `0 f: f 
    filetypes = [(3 j& E$ z2 ?7 |) J+ y/ y* z& b' m 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),; R2 I: {. R* {! N1 S  x! G) Y7 | 
        ('所有文件(*.*)', '*.*')] 
& l4 f5 P+ j' p; e' P6 _) ]2 _1 o  s8 K) n. I 
    file_path = askopenfilename(filetypes=filetypes) 
6 H2 X/ Z- Z& @    return file_path# ~; T) _% v1 i8 X3 n 
 
+ o; @! ~' m; P: H6 N5 f- i2 P6 N; w' N( `/ {# c5 ~4 ] 
def Read_file(file_path):2 g1 I0 l' c( r 
    with open(file_path, mode='rb') as file_object: 
5 ?/ T: t% I$ R4 j  ]5 Q        contents = file_object.read() 
0 ?) C9 ]; b: B# [3 j( `$ d        return contents.rstrip()* W9 Y; j* J& C3 d  B% p 
, p9 d' A' C\" H* L+ K6 m8 V 
7 @4 g2 B' z+ |/ m- g9 N 
def Baidu_ocr(binary): 
/ o3 q8 \+ f- m6 |    url = "https://aip.baidubce.com/oauth/2.0/token?" 
- M! \3 @$ N: r- @    data = {. g. a& }+ B% x 
        "grant_type": "client_credentials",0 e8 v& K: F0 H! G\" ]: N7 K! T% \8 ? 
        "client_id": "",  # 必填8 O7 Z2 v& _4 T, r+ c) d3 Q 
        "client_secret": ""  # 必填5 d# r$ E: u9 @$ V; K3 [/ @1 d 
    } 
/ E5 S* t; Y# E! `0 e5 ?4 L    url += urlencode(data) 
6 x0 {: n+ s. u2 C    response = requests.get(url=url).json(); R4 N( g. A2 j! F4 [% X2 M 
    access_token = response['access_token']- k  K- M8 T; T 
 
7 S+ ^1 d6 m7 s- \9 S; u0 W5 Z    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
: K2 W$ A$ D/ t9 j$ L+ G    post_data = {* V! k' x# }. G 
        "image": base64.b64encode(binary),2 H) q1 w/ ^: z9 G2 |' A+ Z 
        "url": "", 
% M) l- o8 E+ E        "language_type": "CHN_ENG",; J1 C, N  h2 [& p7 ` 
        "detect_direction": "false",5 O) E\" R  X! u$ S 
        "detect_language": "false",) I: n. r3 d4 \+ y( O! K\" F! M 
        "probability": "false"% ?1 ?, i5 w  ^) y 
    } 
* O4 t+ b( z; g# m4 V    response = requests.post(url=url, data=post_data).json() 
0 X/ I/ g* I' t, @  p- Z    num = response['words_result_num'] 
, X% J1 H' l' n- _( k$ t, x    result = ''; l& _7 X- W/ ]) \ 
    for i in range(num): 
, v: }* Z8 E) |# j6 H) w6 p        result += response['words_result'][ i]['words'] + "\r\n" 
( @- V: n3 v$ Y\" |    return result& I$ w7 s5 b- B 
- J, I$ F5 z& f3 [8 j. K) p 
 
\" N9 W( Y9 k\" y& X/ H% v1 I# ~5 j) eif __name__ == "__main__": 
( _& t1 _2 b4 {3 _, w6 z    file_path = Dialog_box() 
\" h4 l# R  s$ W, g- o, \$ R7 w& B    if exists(file_path): 
; A% y; X6 R- O/ K: ?. n        data = Read_file(file_path)# P4 @\" H/ O' r! u, z. `7 V8 \! T 
        print(Baidu_ocr(data))\" G) w. J! U& \5 a1 `1 X( j 
: s, K$ p2 k' d( T. R- M5 k) F9 G

转发自派生社区

Python交流群：1047602540

zan