查看: 2708|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/python0 l\" P\" ~- U) ~7 a
# -*- coding: utf-8 -*-/ L9 F+ o3 {. ]4 g/ X! Q4 K1 r) k
. h\" Z! w& E3 X8 J6 Z+ zimport base642 ~ @# e$ ~3 f1 L |
from os.path import exists
3 j7 S8 w9 Y9 Y$ D3 V% Efrom tkinter import Tk
5 c/ \4 p9 t( |( ]from tkinter.filedialog import askopenfilename
& z, ]\" z& b/ t, b3 Zfrom urllib.parse import urlencode
; e4 Z( E' l. l, R: O3 @1 W$ P' W4 r5 O
# x2 ^& e+ X/ o7 ?# ]- [9 Ximport requests
8 D5 W* d+ [\" D& e( q' X6 b6 Y4 ?/ s/ q! K* m* ?8 c
9 f8 p. h0 i6 M$ Edef Dialog_box():; u! ]# h: f2 B8 S\" }0 q) P6 m
root = Tk()$ e$ F' z4 I9 _& N' K+ }9 j
root.withdraw()1 [; O7 Y! L7 J5 {' y- T
4 W4 P5 ~' z% z6 S) _( D
filetypes = [(
o- F# E8 ]9 H5 z '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
$ A0 r; j2 {$ K- T+ d4 |) A0 B ('所有文件(*.*)', '*.*')]
+ P. x+ ^3 d. }8 O8 t1 G! i2 s+ J$ l9 P# F$ \9 m2 Z% @
file_path = askopenfilename(filetypes=filetypes)
4 f. k6 @/ a5 z4 m- r7 w return file_path$ e0 v6 l$ g. A, e: {& R
3 h8 ?\" I! x, |2 A6 N! Q) Z
$ y9 u\" y# y0 U7 G\" j1 m( H
def Read_file(file_path):9 O# j' v* F7 |9 `
with open(file_path, mode='rb') as file_object:
+ C( y' [3 l' b2 w: m$ e n- b contents = file_object.read()
$ d3 @( W2 [+ F% W return contents.rstrip()$ k5 }* e Z$ h9 G
- C; ^6 c6 h; w7 g( [2 m
' v Z7 q }2 e1 N
def Baidu_ocr(binary):, i5 }! W7 q8 \0 b& H8 d
url = "https://aip.baidubce.com/oauth/2.0/token?"% P% T2 @8 G* Z5 K( h2 `$ ]1 e
data = {
7 G8 F$ ]\" s( v+ n- m\" ~& U "grant_type": "client_credentials",/ w' S\" F/ H\" m8 z, j/ R5 _
"client_id": "", # 必填
5 R\" G/ z! X, j; U! N/ z8 }% T "client_secret": "" # 必填
# b) W r9 g* s, a2 I3 H- M; z+ u }0 h+ c+ V( d5 ?, S9 K a
url += urlencode(data)! @2 @6 L$ Q' ~* @\" w
response = requests.get(url=url).json()
, p0 S0 E8 H+ K# U access_token = response['access_token']
' o, U8 j' P$ d7 [
' [( ^6 E: i0 v- c+ \& V url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token
% ? Y* K7 e4 ]6 r9 ^/ `* y! B4 N post_data = {2 W) M3 Z( z1 |- `2 @+ N! ]8 W
"image": base64.b64encode(binary),
; Z: o5 i5 Y9 v "url": "",. r1 q6 f0 V( \! z4 n
"language_type": "CHN_ENG",
& ^\" |( Y4 S6 g `! w "detect_direction": "false",) Q1 ] i( P2 \; u% Q3 s
"detect_language": "false",3 ~3 H5 R6 \& U8 y E, j
"probability": "false"5 m5 _4 A3 K. L9 c+ n0 p
}' z0 U: r9 d; Y/ B6 V
response = requests.post(url=url, data=post_data).json()
% E' \8 U1 {& f; Z num = response['words_result_num']
5 t6 J' ]0 b* j# b result = ''( i/ A1 L: m. T* d6 M2 g
for i in range(num):7 F9 F0 L- x( u\" `/ d
result += response['words_result'][ i]['words'] + "\r\n"4 S+ e6 P' o% V7 h
return result( q0 @6 z* b+ h; R' b
6 a\" l# D6 W$ I; Q3 |2 ]. ?% ?/ E
if __name__ == "__main__":( x9 m7 C3 E* x2 I) X\" l5 Q
file_path = Dialog_box()
- O- A# T( c# p) C if exists(file_path):
2 A5 {' F: w# ]+ h2 P! Y data = Read_file(file_path)
s9 {: l ]. F1 _$ _ print(Baidu_ocr(data))# p. R& Z S$ J7 L
: O$ R1 ?, d0 O' W* e# @! R+ c

#!/usr/bin/python0 l\" P\" ~- U) ~7 a 
# -*- coding: utf-8 -*-/ L9 F+ o3 {. ]4 g/ X! Q4 K1 r) k 
 
. h\" Z! w& E3 X8 J6 Z+ zimport base642 ~  @# e$ ~3 f1 L  | 
from os.path import exists 
3 j7 S8 w9 Y9 Y$ D3 V% Efrom tkinter import Tk 
5 c/ \4 p9 t( |( ]from tkinter.filedialog import askopenfilename 
& z, ]\" z& b/ t, b3 Zfrom urllib.parse import urlencode 
; e4 Z( E' l. l, R: O3 @1 W$ P' W4 r5 O 
# x2 ^& e+ X/ o7 ?# ]- [9 Ximport requests 
8 D5 W* d+ [\" D& e( q' X6 b6 Y4 ?/ s/ q! K* m* ?8 c 
 
9 f8 p. h0 i6 M$ Edef Dialog_box():; u! ]# h: f2 B8 S\" }0 q) P6 m 
    root = Tk()$ e$ F' z4 I9 _& N' K+ }9 j 
    root.withdraw()1 [; O7 Y! L7 J5 {' y- T 
4 W4 P5 ~' z% z6 S) _( D 
    filetypes = [( 
  o- F# E8 ]9 H5 z        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
$ A0 r; j2 {$ K- T+ d4 |) A0 B        ('所有文件(*.*)', '*.*')] 
+ P. x+ ^3 d. }8 O8 t1 G! i2 s+ J$ l9 P# F$ \9 m2 Z% @ 
    file_path = askopenfilename(filetypes=filetypes) 
4 f. k6 @/ a5 z4 m- r7 w    return file_path$ e0 v6 l$ g. A, e: {& R 
3 h8 ?\" I! x, |2 A6 N! Q) Z 
$ y9 u\" y# y0 U7 G\" j1 m( H 
def Read_file(file_path):9 O# j' v* F7 |9 ` 
    with open(file_path, mode='rb') as file_object: 
+ C( y' [3 l' b2 w: m$ e  n- b        contents = file_object.read() 
$ d3 @( W2 [+ F% W        return contents.rstrip()$ k5 }* e  Z$ h9 G 
- C; ^6 c6 h; w7 g( [2 m 
' v  Z7 q  }2 e1 N 
def Baidu_ocr(binary):, i5 }! W7 q8 \0 b& H8 d 
    url = "https://aip.baidubce.com/oauth/2.0/token?"% P% T2 @8 G* Z5 K( h2 `$ ]1 e 
    data = { 
7 G8 F$ ]\" s( v+ n- m\" ~& U        "grant_type": "client_credentials",/ w' S\" F/ H\" m8 z, j/ R5 _ 
        "client_id": "",  # 必填 
5 R\" G/ z! X, j; U! N/ z8 }% T        "client_secret": ""  # 必填 
# b) W  r9 g* s, a2 I3 H- M; z+ u    }0 h+ c+ V( d5 ?, S9 K  a 
    url += urlencode(data)! @2 @6 L$ Q' ~* @\" w 
    response = requests.get(url=url).json() 
, p0 S0 E8 H+ K# U    access_token = response['access_token'] 
' o, U8 j' P$ d7 [ 
' [( ^6 E: i0 v- c+ \& V    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token 
% ?  Y* K7 e4 ]6 r9 ^/ `* y! B4 N    post_data = {2 W) M3 Z( z1 |- `2 @+ N! ]8 W 
        "image": base64.b64encode(binary), 
; Z: o5 i5 Y9 v        "url": "",. r1 q6 f0 V( \! z4 n 
        "language_type": "CHN_ENG", 
& ^\" |( Y4 S6 g  `! w        "detect_direction": "false",) Q1 ]  i( P2 \; u% Q3 s 
        "detect_language": "false",3 ~3 H5 R6 \& U8 y  E, j 
        "probability": "false"5 m5 _4 A3 K. L9 c+ n0 p 
    }' z0 U: r9 d; Y/ B6 V 
    response = requests.post(url=url, data=post_data).json() 
% E' \8 U1 {& f; Z    num = response['words_result_num'] 
5 t6 J' ]0 b* j# b    result = ''( i/ A1 L: m. T* d6 M2 g 
    for i in range(num):7 F9 F0 L- x( u\" `/ d 
        result += response['words_result'][ i]['words'] + "\r\n"4 S+ e6 P' o% V7 h 
    return result( q0 @6 z* b+ h; R' b 
 
6 a\" l# D6 W$ I; Q3 |2 ]. ?% ?/ E 
if __name__ == "__main__":( x9 m7 C3 E* x2 I) X\" l5 Q 
    file_path = Dialog_box() 
- O- A# T( c# p) C    if exists(file_path): 
2 A5 {' F: w# ]+ h2 P! Y        data = Read_file(file_path) 
  s9 {: l  ]. F1 _$ _        print(Baidu_ocr(data))# p. R& Z  S$ J7 L 
: O$ R1 ?, d0 O' W* e# @! R+ c

转发自派生社区

Python交流群：1047602540

zan