查看: 2808|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
0 o m) R9 n& j0 l: i# -*- coding: utf-8 -*-+ J6 B( X- a0 s+ v
) z& ]6 |0 f, L\" n L- o
import base64% e0 d( W/ u/ V9 i
from os.path import exists8 c0 a/ V9 {/ w
from tkinter import Tk/ ~7 P2 q! A5 y5 k! n& l6 v% C
from tkinter.filedialog import askopenfilename0 w0 v$ } G2 U) M5 v! f
from urllib.parse import urlencode0 w% _1 S/ w2 i- N( ]2 ~4 s
3 e1 I4 e% U% r9 Himport requests& w\" _1 c5 R1 K5 E; V/ K
$ e/ y2 T0 }( R: u1 j' j5 s) o8 w0 v2 E6 p& v$ R8 T6 F
def Dialog_box():+ f! t/ ? L, \& D) B4 J& |* y' ?
root = Tk()4 q# w/ `- w+ w. F\" w5 A
root.withdraw()
+ q2 E' |$ t5 M! S
v# ^7 g+ m; q/ P filetypes = [(
, {9 @% z. |+ A '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),1 |( b% s4 k\" f% ?- h7 s
('所有文件(*.*)', '*.*')]
9 z9 a3 m7 j3 h7 P; B2 q7 J. ?
0 I7 x Z4 g\" o+ r9 u+ Q file_path = askopenfilename(filetypes=filetypes)1 e3 Z7 }' ~9 u* t- z; b
return file_path3 j! A! I/ T% v
8 ^8 V- Q) j6 q) I+ a
% v: m$ ?& ~3 z* c4 F7 p( |. A/ ?7 p$ Mdef Read_file(file_path):
5 X5 ]5 Y) `9 v. `: u* c with open(file_path, mode='rb') as file_object:+ Y# T\" b t' R& ^\" |. x% K& [3 x ^
contents = file_object.read()
0 E0 F; J* i/ M h\" ]! _' O1 c return contents.rstrip()+ h8 h4 H2 w/ n4 F) H
% H) v9 E' i9 E8 c* y; S\" \1 k; h6 m
6 i( P2 ^& { U6 [# K
def Baidu_ocr(binary):' X E, l6 u k- R% @
url = "https://aip.baidubce.com/oauth/2.0/token?"2 [! g: Z6 O$ n
data = {) L% l; D9 t) {* d
"grant_type": "client_credentials",4 \& Y! \9 n) R& h7 Y# u4 q# J
"client_id": "", # 必填2 z; t9 l$ \8 y# w
"client_secret": "" # 必填/ L. R( |5 S2 W8 P
}
9 c5 k& [; {( [% |! \ w url += urlencode(data)
( ^5 u) {# y( r, l- o; f response = requests.get(url=url).json() ~$ t. z* X _
access_token = response['access_token']2 u: U$ ]# z1 {5 R
. ^4 t# a4 X- ~( [
url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token' {9 P; Z: i; u* h; `/ K% M! A
post_data = {& s t/ C\" ~3 x- ^' q) I
"image": base64.b64encode(binary),
) W1 Z w\" a0 B* p "url": "",, o' h( R$ t# v/ z8 O- B
"language_type": "CHN_ENG",
0 y/ e% [) `1 n/ k' F: v! g "detect_direction": "false",
* g2 S\" V/ r+ C9 a/ B0 ]! n "detect_language": "false",
. m- {& w2 q1 e% z% M: m( B W! q "probability": "false"
' z( u |/ P6 r0 q- \ }! c\" r) H4 I3 ?' @' d5 I4 u
response = requests.post(url=url, data=post_data).json()
4 s8 W) \* W& @, h' i1 h$ w num = response['words_result_num']8 A4 P+ Q0 N6 B; y0 Q1 o/ O
result = ''7 u7 i: p\" U f$ | i' G
for i in range(num):8 J& |3 n4 s! R% ?( p
result += response['words_result'][ i]['words'] + "\r\n". i- I\" m0 C: M: e2 P
return result) u* e) x2 ?; A- T
6 {5 r% @9 f N: `5 G: a# K, Q5 y2 A7 o T( G/ n; f0 K6 g; R
if __name__ == "__main__":/ B, r) i+ X5 T c5 q7 l
file_path = Dialog_box()
% H& Y6 c9 g* V\" }: k if exists(file_path):$ G, A$ y% \ z1 {! O& l% C
data = Read_file(file_path)$ T+ ? J% g: D+ C& R, e
print(Baidu_ocr(data))& w! |# R# m) r+ T; K0 { r
3 w; [, v$ p) i- v

#!/usr/bin/python 
0 o  m) R9 n& j0 l: i# -*- coding: utf-8 -*-+ J6 B( X- a0 s+ v 
) z& ]6 |0 f, L\" n  L- o 
import base64% e0 d( W/ u/ V9 i 
from os.path import exists8 c0 a/ V9 {/ w 
from tkinter import Tk/ ~7 P2 q! A5 y5 k! n& l6 v% C 
from tkinter.filedialog import askopenfilename0 w0 v$ }  G2 U) M5 v! f 
from urllib.parse import urlencode0 w% _1 S/ w2 i- N( ]2 ~4 s 
 
3 e1 I4 e% U% r9 Himport requests& w\" _1 c5 R1 K5 E; V/ K 
 
$ e/ y2 T0 }( R: u1 j' j5 s) o8 w0 v2 E6 p& v$ R8 T6 F 
def Dialog_box():+ f! t/ ?  L, \& D) B4 J& |* y' ? 
    root = Tk()4 q# w/ `- w+ w. F\" w5 A 
    root.withdraw() 
+ q2 E' |$ t5 M! S 
  v# ^7 g+ m; q/ P    filetypes = [( 
, {9 @% z. |+ A        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),1 |( b% s4 k\" f% ?- h7 s 
        ('所有文件(*.*)', '*.*')] 
9 z9 a3 m7 j3 h7 P; B2 q7 J. ? 
0 I7 x  Z4 g\" o+ r9 u+ Q    file_path = askopenfilename(filetypes=filetypes)1 e3 Z7 }' ~9 u* t- z; b 
    return file_path3 j! A! I/ T% v 
 
8 ^8 V- Q) j6 q) I+ a 
% v: m$ ?& ~3 z* c4 F7 p( |. A/ ?7 p$ Mdef Read_file(file_path): 
5 X5 ]5 Y) `9 v. `: u* c    with open(file_path, mode='rb') as file_object:+ Y# T\" b  t' R& ^\" |. x% K& [3 x  ^ 
        contents = file_object.read() 
0 E0 F; J* i/ M  h\" ]! _' O1 c        return contents.rstrip()+ h8 h4 H2 w/ n4 F) H 
% H) v9 E' i9 E8 c* y; S\" \1 k; h6 m 
6 i( P2 ^& {  U6 [# K 
def Baidu_ocr(binary):' X  E, l6 u  k- R% @ 
    url = "https://aip.baidubce.com/oauth/2.0/token?"2 [! g: Z6 O$ n 
    data = {) L% l; D9 t) {* d 
        "grant_type": "client_credentials",4 \& Y! \9 n) R& h7 Y# u4 q# J 
        "client_id": "",  # 必填2 z; t9 l$ \8 y# w 
        "client_secret": ""  # 必填/ L. R( |5 S2 W8 P 
    } 
9 c5 k& [; {( [% |! \  w    url += urlencode(data) 
( ^5 u) {# y( r, l- o; f    response = requests.get(url=url).json()  ~$ t. z* X  _ 
    access_token = response['access_token']2 u: U$ ]# z1 {5 R 
. ^4 t# a4 X- ~( [ 
    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token' {9 P; Z: i; u* h; `/ K% M! A 
    post_data = {& s  t/ C\" ~3 x- ^' q) I 
        "image": base64.b64encode(binary), 
) W1 Z  w\" a0 B* p        "url": "",, o' h( R$ t# v/ z8 O- B 
        "language_type": "CHN_ENG", 
0 y/ e% [) `1 n/ k' F: v! g        "detect_direction": "false", 
* g2 S\" V/ r+ C9 a/ B0 ]! n        "detect_language": "false", 
. m- {& w2 q1 e% z% M: m( B  W! q        "probability": "false" 
' z( u  |/ P6 r0 q- \    }! c\" r) H4 I3 ?' @' d5 I4 u 
    response = requests.post(url=url, data=post_data).json() 
4 s8 W) \* W& @, h' i1 h$ w    num = response['words_result_num']8 A4 P+ Q0 N6 B; y0 Q1 o/ O 
    result = ''7 u7 i: p\" U  f$ |  i' G 
    for i in range(num):8 J& |3 n4 s! R% ?( p 
        result += response['words_result'][ i]['words'] + "\r\n". i- I\" m0 C: M: e2 P 
    return result) u* e) x2 ?; A- T 
 
6 {5 r% @9 f  N: `5 G: a# K, Q5 y2 A7 o  T( G/ n; f0 K6 g; R 
if __name__ == "__main__":/ B, r) i+ X5 T  c5 q7 l 
    file_path = Dialog_box() 
% H& Y6 c9 g* V\" }: k    if exists(file_path):$ G, A$ y% \  z1 {! O& l% C 
        data = Read_file(file_path)$ T+ ?  J% g: D+ C& R, e 
        print(Baidu_ocr(data))& w! |# R# m) r+ T; K0 {  r 
3 w; [, v$ p) i- v

转发自派生社区

Python交流群：1047602540

zan