查看: 2847|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
7 O: E2 R% o- L; a4 r; o# -*- coding: utf-8 -*-
+ U. J8 v, _) T& x, a% R$ [5 r: ~3 V
( ]5 m2 Q0 b6 H1 i) Iimport base64
- Q: v7 y! v; H8 C( yfrom os.path import exists1 X3 ^! R3 @) n! D! t& R( t
from tkinter import Tk: W- M+ a( G8 N5 S, T: [
from tkinter.filedialog import askopenfilename
/ U- m' K5 W: s# o: |1 Vfrom urllib.parse import urlencode
- V) U# J' L1 Y\" M3 ]0 A) T
4 l0 L/ e1 R1 ?( D* I* bimport requests& A! k4 i+ M! [2 b1 L! s$ P
; k+ ?6 m, y& G# @( L
9 [# i3 z& k0 S. }3 `( H0 X
def Dialog_box():
5 `5 M$ K: e9 V) F* \ root = Tk()% y+ ?: a# ^3 a+ i( E2 O\" R1 |
root.withdraw()
/ X' }: t; I. Q; u7 _9 L' f2 t) N+ ]* D, j( @5 k\" ?* n3 ^
filetypes = [(- @+ j( u2 i/ X, U* f8 L
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
4 R2 p: Q( Y$ O\" y ('所有文件(*.*)', '*.*')]
* h# Y. `9 S- K8 h. f3 H* i7 j
- @: p* e3 `' Y\" t2 h& Y# O) @ file_path = askopenfilename(filetypes=filetypes)
; m0 _! a& n- [! ]3 Q- p return file_path
6 r. K! v5 N- [% z\" `% V
9 `5 T' C$ z% ~# j$ F
4 ?8 h. f) ?4 m& E+ h3 Fdef Read_file(file_path): V- N: P$ _4 V! ~- L7 g
with open(file_path, mode='rb') as file_object:5 ?* [* c% X4 h. s8 t
contents = file_object.read()1 P: T& C3 S& S
return contents.rstrip()
$ L. e* \7 J( s T: b: Q; u2 V
3 A( Z( A9 O. F1 ]6 T i$ I4 B# L' h9 U9 y
def Baidu_ocr(binary):& \: o3 g1 \- x
url = "https://aip.baidubce.com/oauth/2.0/token?"
: x9 t6 \/ J9 l data = {
% @3 T' D% W' w0 l "grant_type": "client_credentials",! E& S# u3 d6 k$ M
"client_id": "", # 必填; N$ U8 ] A% E; }4 S\" d
"client_secret": "" # 必填3 e0 W% q9 ^3 R9 L8 E. p
}
4 i* f. B& ] c url += urlencode(data)1 Z u# w! a. W! \ N, _# a- c% y
response = requests.get(url=url).json()9 R# A6 h1 y* q/ ?
access_token = response['access_token']% L\" N7 r4 P' b5 _. B6 \; B
4 M* b: ?( j. \9 p& K. y# P\" X url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token% X\" E4 Z+ j3 k$ T) Z
post_data = {% Y* I9 E- u5 j\" l' e/ ?
"image": base64.b64encode(binary),
3 ?9 Y& _! `6 ?% R f' ^) G "url": "",. M/ {; N2 h8 Z\" D7 h( i
"language_type": "CHN_ENG",; G. R. Y, f8 L7 V: G
"detect_direction": "false",# S+ u3 U' K2 w6 o/ s\" y
"detect_language": "false",
3 _ P7 O% V h( [2 _ "probability": "false"
$ T5 U7 z. G# h2 h% M }
\" P5 l Q/ l/ m- O% B) Y response = requests.post(url=url, data=post_data).json()
6 B+ E1 b5 m\" H( i num = response['words_result_num']4 |/ f% E4 C8 j0 x- v
result = ''* w) \; Q$ g7 K9 e
for i in range(num):
* t$ H- A. ]7 U$ h! y W# b( w result += response['words_result'][ i]['words'] + "\r\n"
0 S K$ |9 T% g6 y return result
; A; r$ f$ d3 R' \' b0 G% e+ V
' y' `$ g6 U v$ X+ U- D4 @) i* G @% p\" I9 `- ?+ F# y
if __name__ == "__main__":
! Q8 n1 k5 i8 k file_path = Dialog_box()
+ }& l9 A/ f x/ O0 l if exists(file_path):
( `: H5 ~ m\" v6 C4 B, |7 Z# G data = Read_file(file_path)
% n+ W( Z4 W! T# d- ?+ I. B, \ print(Baidu_ocr(data)): R9 \% Q* ?, q
9 p s+ i3 j+ g) B- _( h0 U

#!/usr/bin/python 
7 O: E2 R% o- L; a4 r; o# -*- coding: utf-8 -*- 
+ U. J8 v, _) T& x, a% R$ [5 r: ~3 V 
( ]5 m2 Q0 b6 H1 i) Iimport base64 
- Q: v7 y! v; H8 C( yfrom os.path import exists1 X3 ^! R3 @) n! D! t& R( t 
from tkinter import Tk: W- M+ a( G8 N5 S, T: [ 
from tkinter.filedialog import askopenfilename 
/ U- m' K5 W: s# o: |1 Vfrom urllib.parse import urlencode 
- V) U# J' L1 Y\" M3 ]0 A) T 
4 l0 L/ e1 R1 ?( D* I* bimport requests& A! k4 i+ M! [2 b1 L! s$ P 
; k+ ?6 m, y& G# @( L 
9 [# i3 z& k0 S. }3 `( H0 X 
def Dialog_box(): 
5 `5 M$ K: e9 V) F* \    root = Tk()% y+ ?: a# ^3 a+ i( E2 O\" R1 | 
    root.withdraw() 
/ X' }: t; I. Q; u7 _9 L' f2 t) N+ ]* D, j( @5 k\" ?* n3 ^ 
    filetypes = [(- @+ j( u2 i/ X, U* f8 L 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
4 R2 p: Q( Y$ O\" y        ('所有文件(*.*)', '*.*')] 
* h# Y. `9 S- K8 h. f3 H* i7 j 
- @: p* e3 `' Y\" t2 h& Y# O) @    file_path = askopenfilename(filetypes=filetypes) 
; m0 _! a& n- [! ]3 Q- p    return file_path 
6 r. K! v5 N- [% z\" `% V 
9 `5 T' C$ z% ~# j$ F 
4 ?8 h. f) ?4 m& E+ h3 Fdef Read_file(file_path):  V- N: P$ _4 V! ~- L7 g 
    with open(file_path, mode='rb') as file_object:5 ?* [* c% X4 h. s8 t 
        contents = file_object.read()1 P: T& C3 S& S 
        return contents.rstrip() 
$ L. e* \7 J( s  T: b: Q; u2 V 
3 A( Z( A9 O. F1 ]6 T  i$ I4 B# L' h9 U9 y 
def Baidu_ocr(binary):& \: o3 g1 \- x 
    url = "https://aip.baidubce.com/oauth/2.0/token?" 
: x9 t6 \/ J9 l    data = { 
% @3 T' D% W' w0 l        "grant_type": "client_credentials",! E& S# u3 d6 k$ M 
        "client_id": "",  # 必填; N$ U8 ]  A% E; }4 S\" d 
        "client_secret": ""  # 必填3 e0 W% q9 ^3 R9 L8 E. p 
    } 
4 i* f. B& ]  c    url += urlencode(data)1 Z  u# w! a. W! \  N, _# a- c% y 
    response = requests.get(url=url).json()9 R# A6 h1 y* q/ ? 
    access_token = response['access_token']% L\" N7 r4 P' b5 _. B6 \; B 
 
4 M* b: ?( j. \9 p& K. y# P\" X    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token% X\" E4 Z+ j3 k$ T) Z 
    post_data = {% Y* I9 E- u5 j\" l' e/ ? 
        "image": base64.b64encode(binary), 
3 ?9 Y& _! `6 ?% R  f' ^) G        "url": "",. M/ {; N2 h8 Z\" D7 h( i 
        "language_type": "CHN_ENG",; G. R. Y, f8 L7 V: G 
        "detect_direction": "false",# S+ u3 U' K2 w6 o/ s\" y 
        "detect_language": "false", 
3 _  P7 O% V  h( [2 _        "probability": "false" 
$ T5 U7 z. G# h2 h% M    } 
\" P5 l  Q/ l/ m- O% B) Y    response = requests.post(url=url, data=post_data).json() 
6 B+ E1 b5 m\" H( i    num = response['words_result_num']4 |/ f% E4 C8 j0 x- v 
    result = ''* w) \; Q$ g7 K9 e 
    for i in range(num): 
* t$ H- A. ]7 U$ h! y  W# b( w        result += response['words_result'][ i]['words'] + "\r\n" 
0 S  K$ |9 T% g6 y    return result 
; A; r$ f$ d3 R' \' b0 G% e+ V 
' y' `$ g6 U  v$ X+ U- D4 @) i* G  @% p\" I9 `- ?+ F# y 
if __name__ == "__main__": 
! Q8 n1 k5 i8 k    file_path = Dialog_box() 
+ }& l9 A/ f  x/ O0 l    if exists(file_path): 
( `: H5 ~  m\" v6 C4 B, |7 Z# G        data = Read_file(file_path) 
% n+ W( Z4 W! T# d- ?+ I. B, \        print(Baidu_ocr(data)): R9 \% Q* ?, q 
 
9 p  s+ i3 j+ g) B- _( h0 U

转发自派生社区

Python交流群：1047602540

zan