查看: 2775|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: K8 g/ L2 x% K0 t. `, A0 r
# -*- coding: utf-8 -*-
q' Z% M! A( o' C1 R. b! v: N- H; W$ Z6 y. G
import base64
* |, E\" Q: }\" Ffrom os.path import exists
, U- Y- k5 G9 k. S% J5 N# Mfrom tkinter import Tk
1 Z3 l' C# Q4 W6 z! I) X* wfrom tkinter.filedialog import askopenfilename4 F/ J5 L0 p5 K- Y ~+ |9 _/ p$ A8 o
from urllib.parse import urlencode# \& `! q\" `& \
# ]% [- V6 H! ?
import requests
! y* ~$ ^4 ]2 B8 m3 Y
- e% p+ x& @$ _. {8 o/ u( L* Y) e; Z8 M. r% P; O) z% w
def Dialog_box():
3 {0 l. M7 C% K: B' l; ] root = Tk()
( O) C) E6 @( s! V root.withdraw()# H- J4 @$ ^- s; m
4 P' X8 m0 W2 q4 c, O0 _ filetypes = [(4 E! P$ h z- X1 s. e% T0 r0 s4 u
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),/ A M5 ~- Z) p* x
('所有文件(*.*)', '*.*')] H# C, z\" t* f `
- X1 D6 o9 N, U8 ?/ L2 N Q
file_path = askopenfilename(filetypes=filetypes)$ X\" W7 \9 \' v
return file_path1 y, p( T: c( S0 u. o1 J. O) k
, `' X6 I+ Y9 X\" U: }8 |* c& e
4 X# r; ~2 ^0 Fdef Read_file(file_path):( G' J0 A! e# d2 H( f\" \& S$ N
with open(file_path, mode='rb') as file_object:
8 T5 L5 y\" N: {1 {\" H contents = file_object.read(), y2 \ [4 ~8 ~, b\" h- L
return contents.rstrip()\" \% H6 k6 \7 M0 K. z
+ K/ V1 M/ Z- m1 A
6 N; M# y6 A! n% G& A% Zdef Baidu_ocr(binary):
2 V* W! I! |. |/ o( {% T/ h4 f url = "https://aip.baidubce.com/oauth/2.0/token?"
! A0 ]2 @6 e0 s$ \ data = {
/ f4 A3 j4 `# Z "grant_type": "client_credentials",; ^8 q9 m/ k' g* E- f. x- P
"client_id": "", # 必填 I1 t: @0 J1 t6 a' u6 o
"client_secret": "" # 必填' @. Z2 S+ r3 e& r4 i& }2 `
}% e# j Y7 p\" r* K- ?: \5 c1 a
url += urlencode(data)3 ~# U, N/ H6 j6 T
response = requests.get(url=url).json()
# p( c! ~' `! P8 A z access_token = response['access_token']
( W$ b. _! Q. b- J- }, ]- C/ f; P- R8 w y+ v% b; h/ K
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
2 V9 q6 v F3 l e8 B post_data = {. b' v- O/ _4 g. ?5 g\" s5 S2 O* P
"image": base64.b64encode(binary),0 Z- z\" T, } H3 K
"url": "",+ C$ N- s+ F9 I5 A/ z( e
"language_type": "CHN_ENG",+ \0 H2 j$ W- \9 Y* ?/ o
"detect_direction": "false",6 S5 `! X: Q* Q( T; ^9 i
"detect_language": "false",
% d6 o( @. H0 Z2 I V, [ z' M "probability": "false"7 n' C- B, M* B
}+ f5 z0 r$ M& \7 d- Q
response = requests.post(url=url, data=post_data).json()
1 p+ {0 C) L* R num = response['words_result_num']
6 d8 a5 N4 z# ~$ A, q result = ''
- u7 n2 {/ k\" o. X1 O8 }\" R' s for i in range(num):% f8 n/ p5 K7 u4 { j2 N; U r
result += response['words_result'][ i]['words'] + "\r\n"
6 N% K0 J9 ` |. M4 ?- h return result
0 n$ a+ `6 ~: k+ ?( U2 `% M- N\" W' H8 e# B) W7 T3 x
! k3 U3 Z/ M$ b( H3 ~& W- hif __name__ == "__main__":
7 |1 o5 K! \' G. M* E0 t file_path = Dialog_box()# H& M/ ~- E\" Y! Q, Q/ ^8 A8 I9 n
if exists(file_path):0 \ t( S6 V! B7 \ y c9 m
data = Read_file(file_path)2 c( ?\" n4 l: f% V0 v8 T+ v
print(Baidu_ocr(data))7 m# _, m& u. c
5 M) z/ Y% P6 u' {\" ]

#!/usr/bin/python: K8 g/ L2 x% K0 t. `, A0 r 
# -*- coding: utf-8 -*- 
  q' Z% M! A( o' C1 R. b! v: N- H; W$ Z6 y. G 
import base64 
* |, E\" Q: }\" Ffrom os.path import exists 
, U- Y- k5 G9 k. S% J5 N# Mfrom tkinter import Tk 
1 Z3 l' C# Q4 W6 z! I) X* wfrom tkinter.filedialog import askopenfilename4 F/ J5 L0 p5 K- Y  ~+ |9 _/ p$ A8 o 
from urllib.parse import urlencode# \& `! q\" `& \ 
# ]% [- V6 H! ? 
import requests 
! y* ~$ ^4 ]2 B8 m3 Y 
- e% p+ x& @$ _. {8 o/ u( L* Y) e; Z8 M. r% P; O) z% w 
def Dialog_box(): 
3 {0 l. M7 C% K: B' l; ]    root = Tk() 
( O) C) E6 @( s! V    root.withdraw()# H- J4 @$ ^- s; m 
 
4 P' X8 m0 W2 q4 c, O0 _    filetypes = [(4 E! P$ h  z- X1 s. e% T0 r0 s4 u 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),/ A  M5 ~- Z) p* x 
        ('所有文件(*.*)', '*.*')]  H# C, z\" t* f  ` 
- X1 D6 o9 N, U8 ?/ L2 N  Q 
    file_path = askopenfilename(filetypes=filetypes)$ X\" W7 \9 \' v 
    return file_path1 y, p( T: c( S0 u. o1 J. O) k 
 
, `' X6 I+ Y9 X\" U: }8 |* c& e 
4 X# r; ~2 ^0 Fdef Read_file(file_path):( G' J0 A! e# d2 H( f\" \& S$ N 
    with open(file_path, mode='rb') as file_object: 
8 T5 L5 y\" N: {1 {\" H        contents = file_object.read(), y2 \  [4 ~8 ~, b\" h- L 
        return contents.rstrip()\" \% H6 k6 \7 M0 K. z 
+ K/ V1 M/ Z- m1 A 
 
6 N; M# y6 A! n% G& A% Zdef Baidu_ocr(binary): 
2 V* W! I! |. |/ o( {% T/ h4 f    url = "https://aip.baidubce.com/oauth/2.0/token?" 
! A0 ]2 @6 e0 s$ \    data = { 
/ f4 A3 j4 `# Z        "grant_type": "client_credentials",; ^8 q9 m/ k' g* E- f. x- P 
        "client_id": "",  # 必填  I1 t: @0 J1 t6 a' u6 o 
        "client_secret": ""  # 必填' @. Z2 S+ r3 e& r4 i& }2 ` 
    }% e# j  Y7 p\" r* K- ?: \5 c1 a 
    url += urlencode(data)3 ~# U, N/ H6 j6 T 
    response = requests.get(url=url).json() 
# p( c! ~' `! P8 A  z    access_token = response['access_token'] 
( W$ b. _! Q. b- J- }, ]- C/ f; P- R8 w  y+ v% b; h/ K 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
2 V9 q6 v  F3 l  e8 B    post_data = {. b' v- O/ _4 g. ?5 g\" s5 S2 O* P 
        "image": base64.b64encode(binary),0 Z- z\" T, }  H3 K 
        "url": "",+ C$ N- s+ F9 I5 A/ z( e 
        "language_type": "CHN_ENG",+ \0 H2 j$ W- \9 Y* ?/ o 
        "detect_direction": "false",6 S5 `! X: Q* Q( T; ^9 i 
        "detect_language": "false", 
% d6 o( @. H0 Z2 I  V, [  z' M        "probability": "false"7 n' C- B, M* B 
    }+ f5 z0 r$ M& \7 d- Q 
    response = requests.post(url=url, data=post_data).json() 
1 p+ {0 C) L* R    num = response['words_result_num'] 
6 d8 a5 N4 z# ~$ A, q    result = '' 
- u7 n2 {/ k\" o. X1 O8 }\" R' s    for i in range(num):% f8 n/ p5 K7 u4 {  j2 N; U  r 
        result += response['words_result'][ i]['words'] + "\r\n" 
6 N% K0 J9 `  |. M4 ?- h    return result 
0 n$ a+ `6 ~: k+ ?( U2 `% M- N\" W' H8 e# B) W7 T3 x 
 
! k3 U3 Z/ M$ b( H3 ~& W- hif __name__ == "__main__": 
7 |1 o5 K! \' G. M* E0 t    file_path = Dialog_box()# H& M/ ~- E\" Y! Q, Q/ ^8 A8 I9 n 
    if exists(file_path):0 \  t( S6 V! B7 \  y  c9 m 
        data = Read_file(file_path)2 c( ?\" n4 l: f% V0 v8 T+ v 
        print(Baidu_ocr(data))7 m# _, m& u. c 
5 M) z/ Y% P6 u' {\" ]

转发自派生社区

Python交流群：1047602540

zan