查看: 2773|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$ I4 o! B/ y& L' W) O) C/ C
# -*- coding: utf-8 -*-
; c n0 Q9 Q9 B2 m5 o\" S2 ]4 C h8 F9 `8 U
import base64$ K9 e0 C _: K% |) l4 x
from os.path import exists
\" A7 s' f& \2 P* m+ mfrom tkinter import Tk
* Z4 Z4 n; t& B. E8 G$ a\" Tfrom tkinter.filedialog import askopenfilename( k) y; |- R\" a5 S
from urllib.parse import urlencode
6 [+ O3 M$ X3 T% v2 G4 M% S6 |% k( l7 O+ e, x% f2 [
import requests
* v( M2 j+ l8 B2 C0 {9 O
2 A w8 d& g& }( ^& N
, _ ?$ Y, r2 k; U1 _0 h: adef Dialog_box():
1 j2 A2 b& i7 { root = Tk()8 t5 u! _0 [5 \& D
root.withdraw()- L0 ?! h( B8 F+ f3 y( `0 F* H\" b
& D; r8 ^\" v\" z; l filetypes = [(
% x2 f; F* ~( M! u3 r, h '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),% B- m' j9 W' ?% q7 J u, I; i
('所有文件(*.*)', '*.*')]
- g8 o; ]6 V$ ?1 Q\" s- a5 e0 \- {1 u6 M
file_path = askopenfilename(filetypes=filetypes)5 N4 f. a( x3 @) i% n/ v& ~; n
return file_path
/ ?1 J\" o/ @! K2 P- K# G+ I3 C0 o! E6 S
* ~- Z$ i2 D3 Q1 d2 P/ Edef Read_file(file_path):3 Z5 ~/ u9 m; R5 r
with open(file_path, mode='rb') as file_object:( j; D# ^$ ~; }, z& }0 D
contents = file_object.read()' N: G- a. R; G8 N$ h# L/ `! b
return contents.rstrip()
) @$ _) o9 y\" H3 g4 E# K
* d7 [1 B$ l9 P! G& r\" f; O0 l y5 V
def Baidu_ocr(binary):0 n. ^4 W3 S, @) |\" z
url = "https://aip.baidubce.com/oauth/2.0/token?"$ n) E& W7 _1 x% @1 ]* k
data = {' T$ o4 [1 p q6 n1 X0 H! o# \
"grant_type": "client_credentials",
W- h+ f1 ?\" H: z. v& F+ d/ s "client_id": "", # 必填4 Q9 s3 R- p& X
"client_secret": "" # 必填
\" X- U% f. e) i+ _1 w2 I! X }. t3 j: D5 g+ t7 @
url += urlencode(data)
% {6 G\" R s. ?\" I response = requests.get(url=url).json()
) H% a# z% T- U access_token = response['access_token']5 i, x- s& @\" u) V
* Z, K; Z/ h! u
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token/ n4 x- b+ f, a5 a: q
post_data = {- _' b6 b, {) ^/ y, w( v
"image": base64.b64encode(binary),
% i8 v7 ^ [' ` Q9 T) k "url": "",
- p+ Q# r! ]4 Q w; Z! ~0 p "language_type": "CHN_ENG",7 o- p. H! A% k) t
"detect_direction": "false",
' y! g, x; G1 b3 k "detect_language": "false",9 w3 S V, Z4 N$ i2 r5 @
"probability": "false"+ r& g; M! k( R/ A
}
2 _% h1 q; H' ? response = requests.post(url=url, data=post_data).json()- M: ]2 ^4 K7 F* s3 b
num = response['words_result_num']
+ T2 l B: c3 `5 m4 k1 Y' w result = ''
' l% S# A( u) Q; x% z1 J$ H for i in range(num):
& g% H$ G# X+ {4 p+ s& L) [ result += response['words_result'][ i]['words'] + "\r\n") B* e; ~& J X
return result
0 U3 [ p( q/ o2 Y
, V! \) r$ k$ G6 B5 D7 x7 e+ Y
0 _' _) M- t Yif __name__ == "__main__":. {3 o* g$ O0 m, Q& l& z* n2 i' H/ k
file_path = Dialog_box()5 I: R$ C0 b, W! b1 Q5 H# Z
if exists(file_path):
3 B' ~. Y' O; W+ p8 U+ H1 |- f data = Read_file(file_path)
& j5 ~( Q; W( j ?! j% B4 y print(Baidu_ocr(data))
1 \: i8 i: M# q
8 s+ p( Q0 q* g/ A& g

#!/usr/bin/python$ I4 o! B/ y& L' W) O) C/ C 
# -*- coding: utf-8 -*- 
; c  n0 Q9 Q9 B2 m5 o\" S2 ]4 C  h8 F9 `8 U 
import base64$ K9 e0 C  _: K% |) l4 x 
from os.path import exists 
\" A7 s' f& \2 P* m+ mfrom tkinter import Tk 
* Z4 Z4 n; t& B. E8 G$ a\" Tfrom tkinter.filedialog import askopenfilename( k) y; |- R\" a5 S 
from urllib.parse import urlencode 
6 [+ O3 M$ X3 T% v2 G4 M% S6 |% k( l7 O+ e, x% f2 [ 
import requests 
* v( M2 j+ l8 B2 C0 {9 O 
2 A  w8 d& g& }( ^& N 
, _  ?$ Y, r2 k; U1 _0 h: adef Dialog_box(): 
1 j2 A2 b& i7 {    root = Tk()8 t5 u! _0 [5 \& D 
    root.withdraw()- L0 ?! h( B8 F+ f3 y( `0 F* H\" b 
 
& D; r8 ^\" v\" z; l    filetypes = [( 
% x2 f; F* ~( M! u3 r, h        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),% B- m' j9 W' ?% q7 J  u, I; i 
        ('所有文件(*.*)', '*.*')] 
- g8 o; ]6 V$ ?1 Q\" s- a5 e0 \- {1 u6 M 
    file_path = askopenfilename(filetypes=filetypes)5 N4 f. a( x3 @) i% n/ v& ~; n 
    return file_path 
/ ?1 J\" o/ @! K2 P- K# G+ I3 C0 o! E6 S 
 
* ~- Z$ i2 D3 Q1 d2 P/ Edef Read_file(file_path):3 Z5 ~/ u9 m; R5 r 
    with open(file_path, mode='rb') as file_object:( j; D# ^$ ~; }, z& }0 D 
        contents = file_object.read()' N: G- a. R; G8 N$ h# L/ `! b 
        return contents.rstrip() 
) @$ _) o9 y\" H3 g4 E# K 
* d7 [1 B$ l9 P! G& r\" f; O0 l  y5 V 
def Baidu_ocr(binary):0 n. ^4 W3 S, @) |\" z 
    url = "https://aip.baidubce.com/oauth/2.0/token?"$ n) E& W7 _1 x% @1 ]* k 
    data = {' T$ o4 [1 p  q6 n1 X0 H! o# \ 
        "grant_type": "client_credentials", 
  W- h+ f1 ?\" H: z. v& F+ d/ s        "client_id": "",  # 必填4 Q9 s3 R- p& X 
        "client_secret": ""  # 必填 
\" X- U% f. e) i+ _1 w2 I! X    }. t3 j: D5 g+ t7 @ 
    url += urlencode(data) 
% {6 G\" R  s. ?\" I    response = requests.get(url=url).json() 
) H% a# z% T- U    access_token = response['access_token']5 i, x- s& @\" u) V 
* Z, K; Z/ h! u 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token/ n4 x- b+ f, a5 a: q 
    post_data = {- _' b6 b, {) ^/ y, w( v 
        "image": base64.b64encode(binary), 
% i8 v7 ^  [' `  Q9 T) k        "url": "", 
- p+ Q# r! ]4 Q  w; Z! ~0 p        "language_type": "CHN_ENG",7 o- p. H! A% k) t 
        "detect_direction": "false", 
' y! g, x; G1 b3 k        "detect_language": "false",9 w3 S  V, Z4 N$ i2 r5 @ 
        "probability": "false"+ r& g; M! k( R/ A 
    } 
2 _% h1 q; H' ?    response = requests.post(url=url, data=post_data).json()- M: ]2 ^4 K7 F* s3 b 
    num = response['words_result_num'] 
+ T2 l  B: c3 `5 m4 k1 Y' w    result = '' 
' l% S# A( u) Q; x% z1 J$ H    for i in range(num): 
& g% H$ G# X+ {4 p+ s& L) [        result += response['words_result'][ i]['words'] + "\r\n") B* e; ~& J  X 
    return result 
0 U3 [  p( q/ o2 Y 
, V! \) r$ k$ G6 B5 D7 x7 e+ Y 
0 _' _) M- t  Yif __name__ == "__main__":. {3 o* g$ O0 m, Q& l& z* n2 i' H/ k 
    file_path = Dialog_box()5 I: R$ C0 b, W! b1 Q5 H# Z 
    if exists(file_path): 
3 B' ~. Y' O; W+ p8 U+ H1 |- f        data = Read_file(file_path) 
& j5 ~( Q; W( j  ?! j% B4 y        print(Baidu_ocr(data)) 
1 \: i8 i: M# q 
8 s+ p( Q0 q* g/ A& g

转发自派生社区

Python交流群：1047602540

zan