查看: 2774|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# M% a9 D9 c! o m) W5 y
# -*- coding: utf-8 -*-
. \& x! ^, F, H% G+ n7 N* }2 |6 V3 L v0 I, [+ y* m\" O
import base645 A y* g1 z8 ^$ P6 [
from os.path import exists
) Z4 h# d- V& k7 cfrom tkinter import Tk
from tkinter.filedialog import askopenfilename
5 M! u! d: R0 e0 S& q7 E$ Jfrom urllib.parse import urlencode! O3 ^9 f6 C' w* ^4 x
$ W h- g2 w* T! `5 E5 t: dimport requests/ r+ ]: s N$ e7 x* n0 m, N% `
+ [4 u2 y7 P, m! K3 g* {# C
\" p- ~\" F* ~+ G+ [: v# t6 ndef Dialog_box():8 M3 D4 u e' h1 w6 k0 h
root = Tk()
|# _, U% {; }/ Y( |/ U root.withdraw()
1 Z\" I# }+ ^$ p( M\" m7 F$ w
5 y* k% k8 T' D$ p, d filetypes = [(
+ g) L. h9 u7 @ @- {* u '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),, i) E! p3 D( x% E, L
('所有文件(*.*)', '*.*')]
8 K; Q5 b* H: W
- A6 X% d0 x3 e: T file_path = askopenfilename(filetypes=filetypes)
: {9 n$ X4 \$ b8 d5 a return file_path; c$ v' H( X' z5 T* C
9 H! r% m* F\" \4 n* G. p
& K\" k! h% i- h4 j& S F7 \def Read_file(file_path):
! o n7 ~6 M\" O q) ]1 B3 L with open(file_path, mode='rb') as file_object:
% y3 D( ], A. g+ R$ [$ ~ contents = file_object.read()$ c, y\" ?, U' S4 Q4 W# g' C: h5 Y
return contents.rstrip()
9 J: e& w# c7 w! P0 l% y9 m( d( P& V
6 ]8 T& f$ ?! L9 ^% M
def Baidu_ocr(binary):
\" B/ E) ?( Y2 H2 t9 r! j& F url = "https://aip.baidubce.com/oauth/2.0/token?" r& D7 r+ m5 x9 K0 k/ }: E2 C9 g* v
data = {
0 j. N, ~5 x t. W "grant_type": "client_credentials",9 T6 L' L/ w) J9 m+ c5 h
"client_id": "", # 必填
, ]4 A1 U- u- a. a "client_secret": "" # 必填$ x# {, }, o4 l
}+ z$ F# h5 P5 O& \
url += urlencode(data)
4 F* C% |3 W% |! I% y. c, n$ ` response = requests.get(url=url).json()
' c\" ~, c* Z# y; K# y- w\" E access_token = response['access_token']; y6 u! Y( y( ~% j
( {1 ^; z F+ D7 c0 x url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
& |& R% {9 N# [! I( J0 w. x7 G+ Z post_data = {2 q, C1 c! G) L\" Y. L* C( p( I
"image": base64.b64encode(binary),; b- p b$ \9 Q
"url": "",4 p6 C# }- i6 ?; K3 i5 {8 G9 l3 ?& H
"language_type": "CHN_ENG",
7 d$ k0 D8 n6 c& G- \/ L. d$ h! v "detect_direction": "false",1 [2 z4 P! I6 Y {, o R, ]
"detect_language": "false",- q0 \8 C- r9 j8 j: g( a$ y
"probability": "false"
7 A9 |+ z, a* {' F }
' O; [3 R9 I+ ] response = requests.post(url=url, data=post_data).json()
/ ]) Z1 d3 F* i' |5 g1 n5 r& `. A num = response['words_result_num']+ Q- C; u8 v( c4 v- A% @8 e
result = ''- ~2 `6 G4 r* \0 l0 a2 C$ }2 D6 K; i
for i in range(num):) I% O4 o: ^1 z1 M% a Z
result += response['words_result'][ i]['words'] + "\r\n"
) \, G; n& J7 s) C8 M) x return result+ x1 G9 {9 V2 d; |
# g2 v\" v6 v1 a5 h1 _0 n% }/ w- T: o; {) m5 j) F
if __name__ == "__main__":
$ p) R) _0 e\" Q S& w$ g file_path = Dialog_box()
. K% [# `! m' y# R if exists(file_path):% I; C9 d: a! u( \
data = Read_file(file_path)
3 `/ y1 D/ H# C, c& U' d print(Baidu_ocr(data))
6 q) l\" \) U7 j: q6 }: k. y0 ~! v& t! P

#!/usr/bin/python# M% a9 D9 c! o  m) W5 y 
# -*- coding: utf-8 -*- 
. \& x! ^, F, H% G+ n7 N* }2 |6 V3 L  v0 I, [+ y* m\" O 
import base645 A  y* g1 z8 ^$ P6 [ 
from os.path import exists 
) Z4 h# d- V& k7 cfrom tkinter import Tk 
7 ]  x3 h% Y% |# f3 |\" P( h. e/ H\" m  Rfrom tkinter.filedialog import askopenfilename 
5 M! u! d: R0 e0 S& q7 E$ Jfrom urllib.parse import urlencode! O3 ^9 f6 C' w* ^4 x 
 
$ W  h- g2 w* T! `5 E5 t: dimport requests/ r+ ]: s  N$ e7 x* n0 m, N% ` 
 
+ [4 u2 y7 P, m! K3 g* {# C 
\" p- ~\" F* ~+ G+ [: v# t6 ndef Dialog_box():8 M3 D4 u  e' h1 w6 k0 h 
    root = Tk() 
  |# _, U% {; }/ Y( |/ U    root.withdraw() 
1 Z\" I# }+ ^$ p( M\" m7 F$ w 
5 y* k% k8 T' D$ p, d    filetypes = [( 
+ g) L. h9 u7 @  @- {* u        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),, i) E! p3 D( x% E, L 
        ('所有文件(*.*)', '*.*')] 
8 K; Q5 b* H: W 
- A6 X% d0 x3 e: T    file_path = askopenfilename(filetypes=filetypes) 
: {9 n$ X4 \$ b8 d5 a    return file_path; c$ v' H( X' z5 T* C 
 
9 H! r% m* F\" \4 n* G. p 
& K\" k! h% i- h4 j& S  F7 \def Read_file(file_path): 
! o  n7 ~6 M\" O  q) ]1 B3 L    with open(file_path, mode='rb') as file_object: 
% y3 D( ], A. g+ R$ [$ ~        contents = file_object.read()$ c, y\" ?, U' S4 Q4 W# g' C: h5 Y 
        return contents.rstrip() 
9 J: e& w# c7 w! P0 l% y9 m( d( P& V 
6 ]8 T& f$ ?! L9 ^% M 
def Baidu_ocr(binary): 
\" B/ E) ?( Y2 H2 t9 r! j& F    url = "https://aip.baidubce.com/oauth/2.0/token?"  r& D7 r+ m5 x9 K0 k/ }: E2 C9 g* v 
    data = { 
0 j. N, ~5 x  t. W        "grant_type": "client_credentials",9 T6 L' L/ w) J9 m+ c5 h 
        "client_id": "",  # 必填 
, ]4 A1 U- u- a. a        "client_secret": ""  # 必填$ x# {, }, o4 l 
    }+ z$ F# h5 P5 O& \ 
    url += urlencode(data) 
4 F* C% |3 W% |! I% y. c, n$ `    response = requests.get(url=url).json() 
' c\" ~, c* Z# y; K# y- w\" E    access_token = response['access_token']; y6 u! Y( y( ~% j 
 
( {1 ^; z  F+ D7 c0 x    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
& |& R% {9 N# [! I( J0 w. x7 G+ Z    post_data = {2 q, C1 c! G) L\" Y. L* C( p( I 
        "image": base64.b64encode(binary),; b- p  b$ \9 Q 
        "url": "",4 p6 C# }- i6 ?; K3 i5 {8 G9 l3 ?& H 
        "language_type": "CHN_ENG", 
7 d$ k0 D8 n6 c& G- \/ L. d$ h! v        "detect_direction": "false",1 [2 z4 P! I6 Y  {, o  R, ] 
        "detect_language": "false",- q0 \8 C- r9 j8 j: g( a$ y 
        "probability": "false" 
7 A9 |+ z, a* {' F    } 
' O; [3 R9 I+ ]    response = requests.post(url=url, data=post_data).json() 
/ ]) Z1 d3 F* i' |5 g1 n5 r& `. A    num = response['words_result_num']+ Q- C; u8 v( c4 v- A% @8 e 
    result = ''- ~2 `6 G4 r* \0 l0 a2 C$ }2 D6 K; i 
    for i in range(num):) I% O4 o: ^1 z1 M% a  Z 
        result += response['words_result'][ i]['words'] + "\r\n" 
) \, G; n& J7 s) C8 M) x    return result+ x1 G9 {9 V2 d; | 
 
# g2 v\" v6 v1 a5 h1 _0 n% }/ w- T: o; {) m5 j) F 
if __name__ == "__main__": 
$ p) R) _0 e\" Q  S& w$ g    file_path = Dialog_box() 
. K% [# `! m' y# R    if exists(file_path):% I; C9 d: a! u( \ 
        data = Read_file(file_path) 
3 `/ y1 D/ H# C, c& U' d        print(Baidu_ocr(data)) 
6 q) l\" \) U7 j: q6 }: k. y0 ~! v& t! P

转发自派生社区

Python交流群：1047602540

zan