查看: 2843|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
) Q4 i$ k/ ^. q! V0 k& U4 J# -*- coding: utf-8 -*-. E4 L1 [% j4 |- k, O, w! s
: E- f\" Z+ ^: z\" D! R
import base64$ B( k+ y! p9 g- g3 p3 d& W
from os.path import exists
& t1 W. q6 [# i, Y+ n1 c1 \( Y: zfrom tkinter import Tk- G6 Y* E! G: }6 ?% G- O9 U L# m
from tkinter.filedialog import askopenfilename% l. i6 X4 C; d+ d7 q
from urllib.parse import urlencode1 s8 O# |9 }2 m8 w
4 c! F( L: c( v, ~; O6 Y5 pimport requests( n8 O9 E5 [& } }
/ w, z+ G! O, w, j
* ^\" o$ W+ Z$ t8 {
def Dialog_box():
5 e$ b4 Y& ]1 \6 I4 |% Z; L) E root = Tk()* v- }- f, ~7 J, u4 ]' z6 W
root.withdraw()% v& p- A' h2 u4 y1 n+ R y
) W* J: _5 y/ | filetypes = [(
7 o\" |; l: l9 M, f% j9 H/ f( } '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),9 ^! w5 i, f* Q# X- t6 q; P
('所有文件(*.*)', '*.*')]
# V7 {* P9 V; Z8 N- y# _* ?2 Y% K' I- Q: y/ u4 G3 @$ K
file_path = askopenfilename(filetypes=filetypes)/ e$ t1 T/ ^4 U& K: b) K. i
return file_path& u/ S( `4 H3 i( b( T8 r
& r, J4 [9 D8 E; i) }4 L
9 i: y/ m/ @0 @\" R
def Read_file(file_path):
! r+ b f: e4 [/ n1 [% E with open(file_path, mode='rb') as file_object:\" B- e; T0 o) z\" ]) F
contents = file_object.read()
$ @3 b, J! U$ @, m9 R* O8 G return contents.rstrip()# i( M0 Y# A3 F
) q0 T3 t% U' a/ {& u0 E5 K5 {/ Q
def Baidu_ocr(binary):
6 T5 {\" {; A, n; A url = "https://aip.baidubce.com/oauth/2.0/token?"/ L! n2 q y, p
data = {
: N1 D) P5 P! Y: b' h. c3 t "grant_type": "client_credentials",
\" N1 f9 R/ |$ u9 e\" i0 T0 W "client_id": "", # 必填
0 L2 h _6 u$ z9 `/ c "client_secret": "" # 必填
9 [* k, ?+ }' B' s }
9 P7 |+ l- [( k0 t url += urlencode(data)
; f. q g3 G. P+ O; h6 u4 P response = requests.get(url=url).json() ~( A/ [+ H( }
access_token = response['access_token']3 a% u. x, Q, ]1 l) k% i% O
/ A# D\" {9 p\" w) v/ w url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
5 A$ q4 Z. P, \+ b. ~+ g+ m post_data = {
) ]9 s' ~7 Y6 N8 g: E/ s "image": base64.b64encode(binary),, Z\" G+ E7 A9 w |# y3 }
"url": "",
, l2 E; J4 e/ E0 Q9 Q$ }; b( ^ "language_type": "CHN_ENG",6 N/ k+ e) n N
"detect_direction": "false",/ N$ ~9 Y$ j4 Q4 z- b
"detect_language": "false",
& h7 f4 T6 v- j8 T0 `! b2 Z "probability": "false"
! m$ t# j% ]% S }; n8 h5 O' b- j
response = requests.post(url=url, data=post_data).json()2 Y- c: \* `8 g, A
num = response['words_result_num']
* X9 l+ E. |+ p result = ''/ }- ?+ V6 E( X: I! w
for i in range(num):
6 Y+ ~3 q8 |9 U$ q$ t result += response['words_result'][ i]['words'] + "\r\n"3 N* |5 @# |1 _- p! B5 ~* w
return result. ]2 A, T8 t\" E) C4 H
+ R) P3 B/ ? N$ b d/ D- X
! e( T# u6 n! j* ?2 }' Hif __name__ == "__main__":- U M5 }: H4 A\" H3 }$ O; q6 X7 b
file_path = Dialog_box()2 M- n7 z3 [! ?% J6 {$ b# U' |
if exists(file_path):: m6 \: u' [' l ?* w
data = Read_file(file_path)
4 ~4 @2 @1 ^- E$ G# n; J; T print(Baidu_ocr(data))
& X( a: \7 L2 S! [: c$ d
! z' l8 ?9 L) t2 Y

#!/usr/bin/python 
) Q4 i$ k/ ^. q! V0 k& U4 J# -*- coding: utf-8 -*-. E4 L1 [% j4 |- k, O, w! s 
: E- f\" Z+ ^: z\" D! R 
import base64$ B( k+ y! p9 g- g3 p3 d& W 
from os.path import exists 
& t1 W. q6 [# i, Y+ n1 c1 \( Y: zfrom tkinter import Tk- G6 Y* E! G: }6 ?% G- O9 U  L# m 
from tkinter.filedialog import askopenfilename% l. i6 X4 C; d+ d7 q 
from urllib.parse import urlencode1 s8 O# |9 }2 m8 w 
 
4 c! F( L: c( v, ~; O6 Y5 pimport requests( n8 O9 E5 [& }  } 
/ w, z+ G! O, w, j 
* ^\" o$ W+ Z$ t8 { 
def Dialog_box(): 
5 e$ b4 Y& ]1 \6 I4 |% Z; L) E    root = Tk()* v- }- f, ~7 J, u4 ]' z6 W 
    root.withdraw()% v& p- A' h2 u4 y1 n+ R  y 
 
) W* J: _5 y/ |    filetypes = [( 
7 o\" |; l: l9 M, f% j9 H/ f( }        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),9 ^! w5 i, f* Q# X- t6 q; P 
        ('所有文件(*.*)', '*.*')] 
# V7 {* P9 V; Z8 N- y# _* ?2 Y% K' I- Q: y/ u4 G3 @$ K 
    file_path = askopenfilename(filetypes=filetypes)/ e$ t1 T/ ^4 U& K: b) K. i 
    return file_path& u/ S( `4 H3 i( b( T8 r 
& r, J4 [9 D8 E; i) }4 L 
9 i: y/ m/ @0 @\" R 
def Read_file(file_path): 
! r+ b  f: e4 [/ n1 [% E    with open(file_path, mode='rb') as file_object:\" B- e; T0 o) z\" ]) F 
        contents = file_object.read() 
$ @3 b, J! U$ @, m9 R* O8 G        return contents.rstrip()# i( M0 Y# A3 F 
 
) q0 T3 t% U' a/ {& u0 E5 K5 {/ Q 
def Baidu_ocr(binary): 
6 T5 {\" {; A, n; A    url = "https://aip.baidubce.com/oauth/2.0/token?"/ L! n2 q  y, p 
    data = { 
: N1 D) P5 P! Y: b' h. c3 t        "grant_type": "client_credentials", 
\" N1 f9 R/ |$ u9 e\" i0 T0 W        "client_id": "",  # 必填 
0 L2 h  _6 u$ z9 `/ c        "client_secret": ""  # 必填 
9 [* k, ?+ }' B' s    } 
9 P7 |+ l- [( k0 t    url += urlencode(data) 
; f. q  g3 G. P+ O; h6 u4 P    response = requests.get(url=url).json()  ~( A/ [+ H( } 
    access_token = response['access_token']3 a% u. x, Q, ]1 l) k% i% O 
 
/ A# D\" {9 p\" w) v/ w    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
5 A$ q4 Z. P, \+ b. ~+ g+ m    post_data = { 
) ]9 s' ~7 Y6 N8 g: E/ s        "image": base64.b64encode(binary),, Z\" G+ E7 A9 w  |# y3 } 
        "url": "", 
, l2 E; J4 e/ E0 Q9 Q$ }; b( ^        "language_type": "CHN_ENG",6 N/ k+ e) n  N 
        "detect_direction": "false",/ N$ ~9 Y$ j4 Q4 z- b 
        "detect_language": "false", 
& h7 f4 T6 v- j8 T0 `! b2 Z        "probability": "false" 
! m$ t# j% ]% S    }; n8 h5 O' b- j 
    response = requests.post(url=url, data=post_data).json()2 Y- c: \* `8 g, A 
    num = response['words_result_num'] 
* X9 l+ E. |+ p    result = ''/ }- ?+ V6 E( X: I! w 
    for i in range(num): 
6 Y+ ~3 q8 |9 U$ q$ t        result += response['words_result'][ i]['words'] + "\r\n"3 N* |5 @# |1 _- p! B5 ~* w 
    return result. ]2 A, T8 t\" E) C4 H 
+ R) P3 B/ ?  N$ b  d/ D- X 
 
! e( T# u6 n! j* ?2 }' Hif __name__ == "__main__":- U  M5 }: H4 A\" H3 }$ O; q6 X7 b 
    file_path = Dialog_box()2 M- n7 z3 [! ?% J6 {$ b# U' | 
    if exists(file_path):: m6 \: u' [' l  ?* w 
        data = Read_file(file_path) 
4 ~4 @2 @1 ^- E$ G# n; J; T        print(Baidu_ocr(data)) 
& X( a: \7 L2 S! [: c$ d 
! z' l8 ?9 L) t2 Y

转发自派生社区

Python交流群：1047602540

zan