查看: 2816|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
3 }. F( I% h! g/ q; {\" l3 ~# -*- coding: utf-8 -*-
* h8 {5 w2 T5 U* F7 g: Y9 x
( V+ {7 V3 |3 r4 H& vimport base642 M& y0 \9 K4 }* x& V' z
from os.path import exists8 m) S0 n/ D- Z. E; P\" o
from tkinter import Tk l: ~$ j' l& l- B9 `$ W7 A1 x+ |( C
from tkinter.filedialog import askopenfilename
: l! s( b1 j& [; E! qfrom urllib.parse import urlencode
% I1 n; x ^( T$ k% C6 ^
/ ?* D/ \4 l5 B\" }import requests
. M8 M8 |7 f' Y% _& N! g0 V2 s* f8 n\" v* b+ p# P2 x
- T \; j$ r( I- \3 N\" H+ Mdef Dialog_box():
# ]6 b# K- k1 H4 o, ^ root = Tk()9 b9 W, c3 F( V N; Z; k) d; P( \; w\" g
root.withdraw()
}1 | [1 r6 d) E8 w7 i; Z7 V8 M# |
filetypes = [(
8 d2 P\" F; k\" i\" ]1 \ '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),: p z# I2 S% Z$ e
('所有文件(*.*)', '*.*')]
8 ` g9 R0 l3 W; Y, V9 ]- p: k
file_path = askopenfilename(filetypes=filetypes)
' X, `\" g ^2 W' F$ I8 t! Y0 S return file_path
; d7 q' @: L9 x\" s$ b/ [. J3 E, N! h/ ?4 G5 ^$ E
* x4 W8 F3 G: G+ _
def Read_file(file_path):/ s. t/ |& q- X4 t+ N. R# ]2 n( g
with open(file_path, mode='rb') as file_object:4 u! G+ U3 }' {3 ]5 P! ~, {# d5 E( ^
contents = file_object.read()
J; `& \) p; O, w return contents.rstrip()
6 l) P1 r\" M3 w. G. G9 ~. C& m) g2 _7 {* ~+ o! E6 @) _# b
+ Z; c$ B2 U5 h' w- X% R
def Baidu_ocr(binary):1 `& j C j: N- ]) {
url = "https://aip.baidubce.com/oauth/2.0/token?"3 H: F/ X# b0 X/ Q+ [8 ^
data = {
2 @6 P2 M) H/ E& O+ v "grant_type": "client_credentials",
, {8 S9 ]( I2 `% u "client_id": "", # 必填5 p s' h, ?/ K) A1 I5 J; o
"client_secret": "" # 必填\" e# C6 p: u* z9 J
}
. h4 u! I. b& X1 C. f1 m, { url += urlencode(data). ^7 J1 M) n5 e6 B/ C ^
response = requests.get(url=url).json()
3 d) i2 ~, e9 c( ]+ x/ s3 N& d! ~ access_token = response['access_token']
/ w! N% k! R4 |5 ?) R
) [! h' G6 [ z8 V9 { url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token# g$ Z5 q9 Z! ]9 h7 x2 y
post_data = {
7 ~* h: e2 L: ?3 ^8 w! w( `: Q "image": base64.b64encode(binary),; B3 S$ g/ }8 b, L2 i. I
"url": "",. k; l4 q9 T2 j4 X& r
"language_type": "CHN_ENG",
- N6 ]% F\" |0 s7 S1 n2 M "detect_direction": "false",$ w6 [- w9 e, q% d7 ^0 V
"detect_language": "false",\" M! G1 _) \7 v0 \. j9 Q
"probability": "false"
6 I# t, Z# g3 ]) V8 B4 E }+ D2 c1 j+ [, A1 [: ^
response = requests.post(url=url, data=post_data).json()
; w\" b5 L+ y& g( D num = response['words_result_num']$ [4 y$ @; s\" T* U7 U' ~4 p* K
result = ''
( A8 H+ ?* S/ p1 j5 r7 d; w for i in range(num):# S) _( @4 b0 Y* W
result += response['words_result'][ i]['words'] + "\r\n"
7 [( ~. H* U\" i: @% L+ k4 y return result. V& `( g\" l- Y3 D& @
9 Y* E\" U: _/ Z( X' D4 \2 p5 ]) x# X' n- Z0 {7 }
if __name__ == "__main__":7 o9 m8 f; m0 N
file_path = Dialog_box()* R3 l$ N5 z+ q9 i
if exists(file_path):+ u v# s+ P3 H6 S9 i5 S
data = Read_file(file_path)
$ ?' p% h; ~) `2 T% q, ~( H print(Baidu_ocr(data))
% ^\" e1 M; x\" G- l! e* w' `# r' y& |
6 G% f& I& g8 e- A$ D+ \

#!/usr/bin/python 
3 }. F( I% h! g/ q; {\" l3 ~# -*- coding: utf-8 -*- 
* h8 {5 w2 T5 U* F7 g: Y9 x 
( V+ {7 V3 |3 r4 H& vimport base642 M& y0 \9 K4 }* x& V' z 
from os.path import exists8 m) S0 n/ D- Z. E; P\" o 
from tkinter import Tk  l: ~$ j' l& l- B9 `$ W7 A1 x+ |( C 
from tkinter.filedialog import askopenfilename 
: l! s( b1 j& [; E! qfrom urllib.parse import urlencode 
% I1 n; x  ^( T$ k% C6 ^ 
/ ?* D/ \4 l5 B\" }import requests 
. M8 M8 |7 f' Y% _& N! g0 V2 s* f8 n\" v* b+ p# P2 x 
 
- T  \; j$ r( I- \3 N\" H+ Mdef Dialog_box(): 
# ]6 b# K- k1 H4 o, ^    root = Tk()9 b9 W, c3 F( V  N; Z; k) d; P( \; w\" g 
    root.withdraw() 
  }1 |  [1 r6 d) E8 w7 i; Z7 V8 M# | 
    filetypes = [( 
8 d2 P\" F; k\" i\" ]1 \        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),: p  z# I2 S% Z$ e 
        ('所有文件(*.*)', '*.*')] 
8 `  g9 R0 l3 W; Y, V9 ]- p: k 
    file_path = askopenfilename(filetypes=filetypes) 
' X, `\" g  ^2 W' F$ I8 t! Y0 S    return file_path 
; d7 q' @: L9 x\" s$ b/ [. J3 E, N! h/ ?4 G5 ^$ E 
* x4 W8 F3 G: G+ _ 
def Read_file(file_path):/ s. t/ |& q- X4 t+ N. R# ]2 n( g 
    with open(file_path, mode='rb') as file_object:4 u! G+ U3 }' {3 ]5 P! ~, {# d5 E( ^ 
        contents = file_object.read() 
  J; `& \) p; O, w        return contents.rstrip() 
6 l) P1 r\" M3 w. G. G9 ~. C& m) g2 _7 {* ~+ o! E6 @) _# b 
+ Z; c$ B2 U5 h' w- X% R 
def Baidu_ocr(binary):1 `& j  C  j: N- ]) { 
    url = "https://aip.baidubce.com/oauth/2.0/token?"3 H: F/ X# b0 X/ Q+ [8 ^ 
    data = { 
2 @6 P2 M) H/ E& O+ v        "grant_type": "client_credentials", 
, {8 S9 ]( I2 `% u        "client_id": "",  # 必填5 p  s' h, ?/ K) A1 I5 J; o 
        "client_secret": ""  # 必填\" e# C6 p: u* z9 J 
    } 
. h4 u! I. b& X1 C. f1 m, {    url += urlencode(data). ^7 J1 M) n5 e6 B/ C  ^ 
    response = requests.get(url=url).json() 
3 d) i2 ~, e9 c( ]+ x/ s3 N& d! ~    access_token = response['access_token'] 
/ w! N% k! R4 |5 ?) R 
) [! h' G6 [  z8 V9 {    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token# g$ Z5 q9 Z! ]9 h7 x2 y 
    post_data = { 
7 ~* h: e2 L: ?3 ^8 w! w( `: Q        "image": base64.b64encode(binary),; B3 S$ g/ }8 b, L2 i. I 
        "url": "",. k; l4 q9 T2 j4 X& r 
        "language_type": "CHN_ENG", 
- N6 ]% F\" |0 s7 S1 n2 M        "detect_direction": "false",$ w6 [- w9 e, q% d7 ^0 V 
        "detect_language": "false",\" M! G1 _) \7 v0 \. j9 Q 
        "probability": "false" 
6 I# t, Z# g3 ]) V8 B4 E    }+ D2 c1 j+ [, A1 [: ^ 
    response = requests.post(url=url, data=post_data).json() 
; w\" b5 L+ y& g( D    num = response['words_result_num']$ [4 y$ @; s\" T* U7 U' ~4 p* K 
    result = '' 
( A8 H+ ?* S/ p1 j5 r7 d; w    for i in range(num):# S) _( @4 b0 Y* W 
        result += response['words_result'][ i]['words'] + "\r\n" 
7 [( ~. H* U\" i: @% L+ k4 y    return result. V& `( g\" l- Y3 D& @ 
 
9 Y* E\" U: _/ Z( X' D4 \2 p5 ]) x# X' n- Z0 {7 } 
if __name__ == "__main__":7 o9 m8 f; m0 N 
    file_path = Dialog_box()* R3 l$ N5 z+ q9 i 
    if exists(file_path):+ u  v# s+ P3 H6 S9 i5 S 
        data = Read_file(file_path) 
$ ?' p% h; ~) `2 T% q, ~( H        print(Baidu_ocr(data)) 
% ^\" e1 M; x\" G- l! e* w' `# r' y& | 
6 G% f& I& g8 e- A$ D+ \

转发自派生社区

Python交流群：1047602540

zan