查看: 2769|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
0 ~/ A( }& J$ [# z# -*- coding: utf-8 -*-
/ T$ h, y- ?( q+ \: v; i0 ?4 U3 | d* i, W. P2 B/ ^
import base64
* U' x, T8 j$ M0 p* A6 j$ b& Gfrom os.path import exists
0 W, `! N2 i\" F( C+ g. z4 S$ Q# Ufrom tkinter import Tk
% W: m; B2 b g1 Y5 Bfrom tkinter.filedialog import askopenfilename9 { @0 i1 ^* h4 S# i# f
from urllib.parse import urlencode/ q/ D( r7 x* d) S& C0 s! i. L1 ~
0 O' n8 r b. [* [! M) N
import requests
$ i7 D. V4 { u% _
8 f0 h& E% F5 L! z8 Z\" k9 I- t6 F9 X! u* A$ r2 P( E
def Dialog_box():$ g, x; D$ P! q. \' o; }+ D
root = Tk()
$ l/ Y: H/ d, S; b: g: q; A0 ?% d root.withdraw()
8 W; _$ }7 } B1 A. v9 T5 r4 \2 p1 y0 o8 Y o7 A
filetypes = [(5 M: R3 {9 k) o: A! W2 p6 |
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),; i( m9 b0 @5 l
('所有文件(*.*)', '*.*')]- o# W+ Z9 I* V2 e5 {/ y, d\" `
0 f' B! n+ A n- h% [ file_path = askopenfilename(filetypes=filetypes)
B$ q+ }% R+ `* ^0 e4 W return file_path7 g4 f2 o7 K5 R) D% A
, ]; h: m\" ~& x4 o9 g5 ]% v. X% I6 R7 O+ }
def Read_file(file_path):
* s% `7 J2 |; h1 e1 Z7 S1 ^5 U9 G- i with open(file_path, mode='rb') as file_object:
9 w8 N+ A5 \7 p1 C4 k' J9 I contents = file_object.read()8 c4 D1 { Y8 S: k
return contents.rstrip()
- Q& O+ s2 {* j. D8 b5 ~
/ `& O/ c, K5 N2 X
% K5 c2 M o; L, r+ c ddef Baidu_ocr(binary):8 p8 l$ _3 v- R) P! T8 y
url = "https://aip.baidubce.com/oauth/2.0/token?"
\" E. r, m& Q! n+ v: c9 s: r data = {' N( v, x! a3 u, J G7 ]
"grant_type": "client_credentials",
$ |: U! L U5 D1 ` "client_id": "", # 必填
\" f' y5 H/ G1 E* S% [ "client_secret": "" # 必填
( ^# f* H& A, F* {+ X J; U1 l }
. ~/ _: ]! w0 B0 J\" L url += urlencode(data)( p2 m9 W! N\" O; z1 [
response = requests.get(url=url).json(): y7 m* J' M$ t0 e) d$ K
access_token = response['access_token']7 _/ X& H) @( Z0 T6 M$ f
7 Y2 x, t3 @- Z: t4 k& L! w S url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token' v, b% Q' t5 B. x' s8 F4 Z6 b
post_data = {
8 J- O. g6 f1 t- i9 U) q0 k- F "image": base64.b64encode(binary),, t! p/ D% o) p; [' U+ u
"url": "",
4 }* a/ O! {8 _, W+ l\" C8 h2 G: H "language_type": "CHN_ENG",# @: H& N5 h, Q' V; w' t
"detect_direction": "false",
8 T; F: b4 }& M5 q7 w; s/ T "detect_language": "false",8 w% k) h4 o7 }9 V
"probability": "false"8 [& K o t* d6 Y. e
}1 G% ~' @6 O% W4 `+ @, p4 J( v
response = requests.post(url=url, data=post_data).json()7 f# U( ~( S) a* E
num = response['words_result_num']9 N( V% k\" N- s\" S' S
result = ''# i+ J2 x$ j* P: Q
for i in range(num):
- a& E0 d) A; ]0 v; r8 l% z result += response['words_result'][ i]['words'] + "\r\n") E0 @# K$ M! w- {8 |3 K& X% k
return result( P0 W5 H$ b7 z3 s/ l
& A- _9 K7 s7 b* {8 n
0 d. ?! s) g8 a( wif __name__ == "__main__":: b) {- n2 M8 J# i* g
file_path = Dialog_box(): v: _* x! u$ S: e3 f
if exists(file_path):
0 `; y. Z- N S( c; |# G' x data = Read_file(file_path)
) e% F* P; p* t% A$ C5 ] print(Baidu_ocr(data)) n, P% p3 A2 ~. F
; M4 z: [9 N, o' `0 i

#!/usr/bin/python 
0 ~/ A( }& J$ [# z# -*- coding: utf-8 -*- 
/ T$ h, y- ?( q+ \: v; i0 ?4 U3 |  d* i, W. P2 B/ ^ 
import base64 
* U' x, T8 j$ M0 p* A6 j$ b& Gfrom os.path import exists 
0 W, `! N2 i\" F( C+ g. z4 S$ Q# Ufrom tkinter import Tk 
% W: m; B2 b  g1 Y5 Bfrom tkinter.filedialog import askopenfilename9 {  @0 i1 ^* h4 S# i# f 
from urllib.parse import urlencode/ q/ D( r7 x* d) S& C0 s! i. L1 ~ 
0 O' n8 r  b. [* [! M) N 
import requests 
$ i7 D. V4 {  u% _ 
8 f0 h& E% F5 L! z8 Z\" k9 I- t6 F9 X! u* A$ r2 P( E 
def Dialog_box():$ g, x; D$ P! q. \' o; }+ D 
    root = Tk() 
$ l/ Y: H/ d, S; b: g: q; A0 ?% d    root.withdraw() 
8 W; _$ }7 }  B1 A. v9 T5 r4 \2 p1 y0 o8 Y  o7 A 
    filetypes = [(5 M: R3 {9 k) o: A! W2 p6 | 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),; i( m9 b0 @5 l 
        ('所有文件(*.*)', '*.*')]- o# W+ Z9 I* V2 e5 {/ y, d\" ` 
 
0 f' B! n+ A  n- h% [    file_path = askopenfilename(filetypes=filetypes) 
  B$ q+ }% R+ `* ^0 e4 W    return file_path7 g4 f2 o7 K5 R) D% A 
 
, ]; h: m\" ~& x4 o9 g5 ]% v. X% I6 R7 O+ } 
def Read_file(file_path): 
* s% `7 J2 |; h1 e1 Z7 S1 ^5 U9 G- i    with open(file_path, mode='rb') as file_object: 
9 w8 N+ A5 \7 p1 C4 k' J9 I        contents = file_object.read()8 c4 D1 {  Y8 S: k 
        return contents.rstrip() 
- Q& O+ s2 {* j. D8 b5 ~ 
/ `& O/ c, K5 N2 X 
% K5 c2 M  o; L, r+ c  ddef Baidu_ocr(binary):8 p8 l$ _3 v- R) P! T8 y 
    url = "https://aip.baidubce.com/oauth/2.0/token?" 
\" E. r, m& Q! n+ v: c9 s: r    data = {' N( v, x! a3 u, J  G7 ] 
        "grant_type": "client_credentials", 
$ |: U! L  U5 D1 `        "client_id": "",  # 必填 
\" f' y5 H/ G1 E* S% [        "client_secret": ""  # 必填 
( ^# f* H& A, F* {+ X  J; U1 l    } 
. ~/ _: ]! w0 B0 J\" L    url += urlencode(data)( p2 m9 W! N\" O; z1 [ 
    response = requests.get(url=url).json(): y7 m* J' M$ t0 e) d$ K 
    access_token = response['access_token']7 _/ X& H) @( Z0 T6 M$ f 
 
7 Y2 x, t3 @- Z: t4 k& L! w  S    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token' v, b% Q' t5 B. x' s8 F4 Z6 b 
    post_data = { 
8 J- O. g6 f1 t- i9 U) q0 k- F        "image": base64.b64encode(binary),, t! p/ D% o) p; [' U+ u 
        "url": "", 
4 }* a/ O! {8 _, W+ l\" C8 h2 G: H        "language_type": "CHN_ENG",# @: H& N5 h, Q' V; w' t 
        "detect_direction": "false", 
8 T; F: b4 }& M5 q7 w; s/ T        "detect_language": "false",8 w% k) h4 o7 }9 V 
        "probability": "false"8 [& K  o  t* d6 Y. e 
    }1 G% ~' @6 O% W4 `+ @, p4 J( v 
    response = requests.post(url=url, data=post_data).json()7 f# U( ~( S) a* E 
    num = response['words_result_num']9 N( V% k\" N- s\" S' S 
    result = ''# i+ J2 x$ j* P: Q 
    for i in range(num): 
- a& E0 d) A; ]0 v; r8 l% z        result += response['words_result'][ i]['words'] + "\r\n") E0 @# K$ M! w- {8 |3 K& X% k 
    return result( P0 W5 H$ b7 z3 s/ l 
& A- _9 K7 s7 b* {8 n 
 
0 d. ?! s) g8 a( wif __name__ == "__main__":: b) {- n2 M8 J# i* g 
    file_path = Dialog_box(): v: _* x! u$ S: e3 f 
    if exists(file_path): 
0 `; y. Z- N  S( c; |# G' x        data = Read_file(file_path) 
) e% F* P; p* t% A$ C5 ]        print(Baidu_ocr(data))  n, P% p3 A2 ~. F 
 
; M4 z: [9 N, o' `0 i

转发自派生社区

Python交流群：1047602540

zan