查看: 2788|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
: J! x. `! h/ Z$ z8 w- n# -*- coding: utf-8 -*-1 @4 ~+ E4 ^. g8 x
; D& c0 i& H- c9 o3 @3 I6 p2 e\" y
import base64
8 j4 R/ ?+ k# P& w+ qfrom os.path import exists5 |7 w7 t/ F) O\" y\" Z
from tkinter import Tk
* h; ]/ c0 w- ]+ O- D5 V. x- B: cfrom tkinter.filedialog import askopenfilename
+ Z# M! T% n8 p4 K) R: C: [/ Ifrom urllib.parse import urlencode( g! g) B# K1 u& c: H9 v
3 T3 G' y$ j! Q% gimport requests
+ `( @5 b9 u& U6 S; S& R, j
7 k% Q, D0 \0 G\" C# p2 `8 b3 Z& z* |( m. T! H
def Dialog_box():1 c8 M8 o$ i5 Z. t% z$ {
root = Tk()# z' j1 E9 U0 K( E2 O# C6 p h
root.withdraw()5 f, f! C% O* F: j3 s; J
0 E1 q\" n2 K# \3 u
filetypes = [(3 o7 A1 N/ X; g& R, |/ m& W2 f
'图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'),
& n$ ^) c, Z0 A4 N9 E7 V ('所有文件(*.*)', '*.*')]\" Q$ v! `7 W9 {
( {. j' r# K$ m\" a
file_path = askopenfilename(filetypes=filetypes)
, z: E- y) F' s$ }0 x. u! O return file_path
1 n\" U/ @! W0 D1 V
: u1 H( f1 t# |& R8 I, Z; M3 ]9 w5 `) r
def Read_file(file_path):
T5 P2 I0 W- b) f$ f) [ with open(file_path, mode='rb') as file_object:
7 c( i) o5 v2 e contents = file_object.read()
# s\" p f0 n0 g. I3 i% G return contents.rstrip()
0 r5 j, W, f6 t8 _6 c, W8 C, e$ U\" t0 s( T$ H3 s
1 _2 t- y$ r$ P& C7 Xdef Baidu_ocr(binary):
9 ]- G% w6 s; H( M url = "https://aip.baidubce.com/oauth/2.0/token?"5 Q6 s. f ?) j: {, B9 P
data = {
4 i {3 R0 U. c9 k "grant_type": "client_credentials",
( K& a: E( @ }) G8 C: d- ^ "client_id": "", # 必填
7 H9 i6 N! n) ^5 v7 c5 A5 Q6 |1 F1 a9 T "client_secret": "" # 必填
0 l8 U- c- M' S8 [1 ?8 K, ^/ M( w }
1 Q- C! E# Y: C2 I! r9 x url += urlencode(data)) @* r8 f) Z7 v0 ^
response = requests.get(url=url).json()
\" X1 s% M: Z7 i, Z' P3 ?6 J access_token = response['access_token']
( Z$ _( d# Q# |3 N& \; {
; I9 d# w8 a\" |& | url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token% Z# N3 T4 g* B% R
post_data = {1 p\" e8 c# Q5 p2 L0 ~9 |
"image": base64.b64encode(binary),% g, Z$ B\" m' I% H! R
"url": "",
' P' B8 N) I\" ] "language_type": "CHN_ENG",$ O\" C6 Q$ s; |5 i( L8 ^
"detect_direction": "false",9 S* \+ y% c7 L
"detect_language": "false",
1 U7 s& z, A3 _# S& A4 h2 S4 d "probability": "false"3 E8 Z3 }* f+ y
}4 P, T; O7 X9 [# k- \$ @- {
response = requests.post(url=url, data=post_data).json()
# A: }# C2 A. z# L. E num = response['words_result_num']
2 _6 y* l& c) s: P, a result = ''7 u7 N\" F l& q7 C$ i
for i in range(num):+ n% j# M5 }/ A7 T) [
result += response['words_result'][ i]['words'] + "\r\n"
4 w$ i+ y0 \: g return result9 V1 Q7 H a s: I2 g7 d% M
0 c; U5 I: R; a/ I/ x0 D* w- a( z i\" Q
if __name__ == "__main__":
( l' ]' r5 c! Q0 i4 Q# X) o file_path = Dialog_box()( [5 y; T2 u R) m, U+ F& Z
if exists(file_path):
9 i\" {7 Z( }/ @ data = Read_file(file_path)\" r9 v1 p! Y8 h4 I- W
print(Baidu_ocr(data))
& ^; _: p0 |$ \% E, d: k- D9 }. Q. F0 [6 H% E7 f0 w

#!/usr/bin/python 
: J! x. `! h/ Z$ z8 w- n# -*- coding: utf-8 -*-1 @4 ~+ E4 ^. g8 x 
; D& c0 i& H- c9 o3 @3 I6 p2 e\" y 
import base64 
8 j4 R/ ?+ k# P& w+ qfrom os.path import exists5 |7 w7 t/ F) O\" y\" Z 
from tkinter import Tk 
* h; ]/ c0 w- ]+ O- D5 V. x- B: cfrom tkinter.filedialog import askopenfilename 
+ Z# M! T% n8 p4 K) R: C: [/ Ifrom urllib.parse import urlencode( g! g) B# K1 u& c: H9 v 
 
3 T3 G' y$ j! Q% gimport requests 
+ `( @5 b9 u& U6 S; S& R, j 
7 k% Q, D0 \0 G\" C# p2 `8 b3 Z& z* |( m. T! H 
def Dialog_box():1 c8 M8 o$ i5 Z. t% z$ { 
    root = Tk()# z' j1 E9 U0 K( E2 O# C6 p  h 
    root.withdraw()5 f, f! C% O* F: j3 s; J 
0 E1 q\" n2 K# \3 u 
    filetypes = [(3 o7 A1 N/ X; g& R, |/ m& W2 f 
        '图片文件(*.jpg;*.jpeg;*.gif;*.png;*.bmp)', '*.jpg;*.jpeg;*.gif;*.png;*.bmp'), 
& n$ ^) c, Z0 A4 N9 E7 V        ('所有文件(*.*)', '*.*')]\" Q$ v! `7 W9 { 
( {. j' r# K$ m\" a 
    file_path = askopenfilename(filetypes=filetypes) 
, z: E- y) F' s$ }0 x. u! O    return file_path 
1 n\" U/ @! W0 D1 V 
: u1 H( f1 t# |& R8 I, Z; M3 ]9 w5 `) r 
def Read_file(file_path): 
  T5 P2 I0 W- b) f$ f) [    with open(file_path, mode='rb') as file_object: 
7 c( i) o5 v2 e        contents = file_object.read() 
# s\" p  f0 n0 g. I3 i% G        return contents.rstrip() 
0 r5 j, W, f6 t8 _6 c, W8 C, e$ U\" t0 s( T$ H3 s 
 
1 _2 t- y$ r$ P& C7 Xdef Baidu_ocr(binary): 
9 ]- G% w6 s; H( M    url = "https://aip.baidubce.com/oauth/2.0/token?"5 Q6 s. f  ?) j: {, B9 P 
    data = { 
4 i  {3 R0 U. c9 k        "grant_type": "client_credentials", 
( K& a: E( @  }) G8 C: d- ^        "client_id": "",  # 必填 
7 H9 i6 N! n) ^5 v7 c5 A5 Q6 |1 F1 a9 T        "client_secret": ""  # 必填 
0 l8 U- c- M' S8 [1 ?8 K, ^/ M( w    } 
1 Q- C! E# Y: C2 I! r9 x    url += urlencode(data)) @* r8 f) Z7 v0 ^ 
    response = requests.get(url=url).json() 
\" X1 s% M: Z7 i, Z' P3 ?6 J    access_token = response['access_token'] 
( Z$ _( d# Q# |3 N& \; { 
; I9 d# w8 a\" |& |    url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic?access_token=" + access_token% Z# N3 T4 g* B% R 
    post_data = {1 p\" e8 c# Q5 p2 L0 ~9 | 
        "image": base64.b64encode(binary),% g, Z$ B\" m' I% H! R 
        "url": "", 
' P' B8 N) I\" ]        "language_type": "CHN_ENG",$ O\" C6 Q$ s; |5 i( L8 ^ 
        "detect_direction": "false",9 S* \+ y% c7 L 
        "detect_language": "false", 
1 U7 s& z, A3 _# S& A4 h2 S4 d        "probability": "false"3 E8 Z3 }* f+ y 
    }4 P, T; O7 X9 [# k- \$ @- { 
    response = requests.post(url=url, data=post_data).json() 
# A: }# C2 A. z# L. E    num = response['words_result_num'] 
2 _6 y* l& c) s: P, a    result = ''7 u7 N\" F  l& q7 C$ i 
    for i in range(num):+ n% j# M5 }/ A7 T) [ 
        result += response['words_result'][ i]['words'] + "\r\n" 
4 w$ i+ y0 \: g    return result9 V1 Q7 H  a  s: I2 g7 d% M 
 
0 c; U5 I: R; a/ I/ x0 D* w- a( z  i\" Q 
if __name__ == "__main__": 
( l' ]' r5 c! Q0 i4 Q# X) o    file_path = Dialog_box()( [5 y; T2 u  R) m, U+ F& Z 
    if exists(file_path): 
9 i\" {7 Z( }/ @        data = Read_file(file_path)\" r9 v1 p! Y8 h4 I- W 
        print(Baidu_ocr(data)) 
& ^; _: p0 |$ \% E, d: k- D9 }. Q. F0 [6 H% E7 f0 w

转发自派生社区

Python交流群：1047602540

zan