查看: 2814|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$ }' U7 c; J1 P7 r
# -*- coding: utf-8 -*-1 P* o0 C0 l9 ?# B' \; Y7 d
. W( v; f+ o/ m$ y1 R\" n
import base64
1 i a+ G6 G2 ~\" x1 R8 L0 c8 J- hfrom os.path import exists
# \: X' f% c0 Y3 u# e7 Zfrom tkinter import Tk, e* h; u8 h+ X- w; L5 {
from tkinter.filedialog import askopenfilename
# d* q; P# t& |: Q ^* Mfrom urllib.parse import urlencode
) e; n3 P6 N% b; k7 U1 h+ Q
: Z; z+ @3 P6 \9 u# F& C t5 e& ~6 fimport requests& x# q- y# R% k7 L/ E% S3 |1 e4 U
& r\" Y( q7 _7 f! L% q
# s$ |6 a( h v& W
def Dialog_box():
& M( y- [& j2 h root = Tk()6 W1 L8 _1 a+ S6 D' ~! A+ Y& U
root.withdraw(); W7 t: v& X- _8 ~
C; p P h/ t6 I7 y filetypes = [(1 u\" ]7 `/ L# x6 ~1 s
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),. ?+ Y& p* y; [& w3 d' W
('所有文件(*.*)', '*.*')]
# B+ s, l7 I; F5 \/ D! {: x' N/ p: J7 k( M) Y0 Z) @- I' ?- |\" e G
file_path = askopenfilename(filetypes=filetypes), P$ ]+ l5 _! p$ U: f: D
return file_path
6 z0 N F( y* f6 p. g [) F# }1 w/ M
7 f# x9 R$ t2 }0 D2 ?
def Read_file(file_path):0 `/ J0 q5 [! D' v3 Y
with open(file_path, mode='rb') as file_object:& M/ K9 u: U6 d\" V R
contents = file_object.read()
% w% D# _- N- W\" y9 o return contents.rstrip()
7 j1 F [$ v# W6 k2 \- B/ k2 t2 e M\" ?- P- R8 W) h& E
- B4 Z( w0 _: s j7 `8 cdef Baidu_ocr(binary):
; C+ H/ x, A' n* k4 N url = "https://aip.baidubce.com/oauth/2.0/token?"
! e/ x' \ b\" p: s7 ^ data = {3 ]! c+ G% \$ P* ?1 x; m
"grant_type": "client_credentials",& I+ o7 b/ i5 t$ b( p. ]
"client_id": "", # 必填
$ s( B2 B1 L5 } "client_secret": "" # 必填
1 y2 @& l, Q2 D+ j* O4 u( N o }# Q1 t8 G. _2 G& I5 A
url += urlencode(data)' v, g* I1 z% E% u$ }3 @
response = requests.get(url=url).json()
. U$ k; v; {/ H- m: r7 n access_token = response['access_token']8 E0 l\" o* e2 M3 p5 b
6 ]' x g& ] U, [ url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token4 m\" N4 g K7 m8 i\" E& s
post_data = {
4 O6 n( V) t+ F c3 w5 f* p4 _ "image": base64.b64encode(binary),
$ R: b, s\" Y( }. v; p$ a/ x "url": "",
& {- Z1 f# k9 H5 Z: ? { "language_type": "CHN_ENG", N+ R6 P; s& K& D' B% k/ k
"detect_direction": "false",& t\" u2 v* a: H5 P& n
"detect_language": "false",
9 j$ K8 h6 F% _. J- J) k "probability": "false"
}4 C+ I- D$ a5 A& m
response = requests.post(url=url, data=post_data).json()
- l2 O! C* G! u G! J* S4 J+ l2 I num = response['words_result_num']# M& g\" g. ]- m; s4 F7 j
result = ''5 {3 V- [; v7 f; f! }' ~1 j4 m. Q
for i in range(num):
) R, j0 Z8 e8 M1 m( Z) p5 i result += response['words_result'][ i]['words'] + "\r\n"+ _/ V6 d( ~/ d
return result
F! d8 g/ s\" N; ?: C- J: a, \- r! a' W: C# L0 S2 D8 t
. }& @$ ~7 P0 d1 G. h
if __name__ == "__main__":
! \. U, }' a! ~8 ?\" i6 P. D file_path = Dialog_box()
4 _3 r5 C\" E4 d1 ^\" k9 r O. P( n if exists(file_path):
6 [. [6 A$ V, Q3 a data = Read_file(file_path)1 u/ |( P2 L% n- W
print(Baidu_ocr(data))
G7 u' o b- }* S\" O7 X: I5 u
6 s/ D9 J7 S5 L/ m

#!/usr/bin/python$ }' U7 c; J1 P7 r 
# -*- coding: utf-8 -*-1 P* o0 C0 l9 ?# B' \; Y7 d 
. W( v; f+ o/ m$ y1 R\" n 
import base64 
1 i  a+ G6 G2 ~\" x1 R8 L0 c8 J- hfrom os.path import exists 
# \: X' f% c0 Y3 u# e7 Zfrom tkinter import Tk, e* h; u8 h+ X- w; L5 { 
from tkinter.filedialog import askopenfilename 
# d* q; P# t& |: Q  ^* Mfrom urllib.parse import urlencode 
) e; n3 P6 N% b; k7 U1 h+ Q 
: Z; z+ @3 P6 \9 u# F& C  t5 e& ~6 fimport requests& x# q- y# R% k7 L/ E% S3 |1 e4 U 
& r\" Y( q7 _7 f! L% q 
# s$ |6 a( h  v& W 
def Dialog_box(): 
& M( y- [& j2 h    root = Tk()6 W1 L8 _1 a+ S6 D' ~! A+ Y& U 
    root.withdraw(); W7 t: v& X- _8 ~ 
 
  C; p  P  h/ t6 I7 y    filetypes = [(1 u\" ]7 `/ L# x6 ~1 s 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),. ?+ Y& p* y; [& w3 d' W 
        ('所有文件(*.*)', '*.*')] 
# B+ s, l7 I; F5 \/ D! {: x' N/ p: J7 k( M) Y0 Z) @- I' ?- |\" e  G 
    file_path = askopenfilename(filetypes=filetypes), P$ ]+ l5 _! p$ U: f: D 
    return file_path 
6 z0 N  F( y* f6 p. g  [) F# }1 w/ M 
7 f# x9 R$ t2 }0 D2 ? 
def Read_file(file_path):0 `/ J0 q5 [! D' v3 Y 
    with open(file_path, mode='rb') as file_object:& M/ K9 u: U6 d\" V  R 
        contents = file_object.read() 
% w% D# _- N- W\" y9 o        return contents.rstrip() 
7 j1 F  [$ v# W6 k2 \- B/ k2 t2 e  M\" ?- P- R8 W) h& E 
 
- B4 Z( w0 _: s  j7 `8 cdef Baidu_ocr(binary): 
; C+ H/ x, A' n* k4 N    url = "https://aip.baidubce.com/oauth/2.0/token?" 
! e/ x' \  b\" p: s7 ^    data = {3 ]! c+ G% \$ P* ?1 x; m 
        "grant_type": "client_credentials",& I+ o7 b/ i5 t$ b( p. ] 
        "client_id": "",  # 必填 
$ s( B2 B1 L5 }        "client_secret": ""  # 必填 
1 y2 @& l, Q2 D+ j* O4 u( N  o    }# Q1 t8 G. _2 G& I5 A 
    url += urlencode(data)' v, g* I1 z% E% u$ }3 @ 
    response = requests.get(url=url).json() 
. U$ k; v; {/ H- m: r7 n    access_token = response['access_token']8 E0 l\" o* e2 M3 p5 b 
 
6 ]' x  g& ]  U, [    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token4 m\" N4 g  K7 m8 i\" E& s 
    post_data = { 
4 O6 n( V) t+ F  c3 w5 f* p4 _        "image": base64.b64encode(binary), 
$ R: b, s\" Y( }. v; p$ a/ x        "url": "", 
& {- Z1 f# k9 H5 Z: ?  {        "language_type": "CHN_ENG",  N+ R6 P; s& K& D' B% k/ k 
        "detect_direction": "false",& t\" u2 v* a: H5 P& n 
        "detect_language": "false", 
9 j$ K8 h6 F% _. J- J) k        "probability": "false" 
  u, K+ h) \% W  v' n  }- _( |  k    }4 C+ I- D$ a5 A& m 
    response = requests.post(url=url, data=post_data).json() 
- l2 O! C* G! u  G! J* S4 J+ l2 I    num = response['words_result_num']# M& g\" g. ]- m; s4 F7 j 
    result = ''5 {3 V- [; v7 f; f! }' ~1 j4 m. Q 
    for i in range(num): 
) R, j0 Z8 e8 M1 m( Z) p5 i        result += response['words_result'][ i]['words'] + "\r\n"+ _/ V6 d( ~/ d 
    return result 
  F! d8 g/ s\" N; ?: C- J: a, \- r! a' W: C# L0 S2 D8 t 
. }& @$ ~7 P0 d1 G. h 
if __name__ == "__main__": 
! \. U, }' a! ~8 ?\" i6 P. D    file_path = Dialog_box() 
4 _3 r5 C\" E4 d1 ^\" k9 r  O. P( n    if exists(file_path): 
6 [. [6 A$ V, Q3 a        data = Read_file(file_path)1 u/ |( P2 L% n- W 
        print(Baidu_ocr(data)) 
  G7 u' o  b- }* S\" O7 X: I5 u 
6 s/ D9 J7 S5 L/ m

转发自派生社区

Python交流群：1047602540

zan