python利用pytesseract 实现本地识别图片文字_Python

python利用pytesseract 实现本地识别图片文字

2021-08-14 00:02凹凸曼大人 Python

这篇文章主要介绍了python利用pytesseract 实现本地识别图片文字，帮助大家更好的理解和使用python，感兴趣的朋友可以了解下

				?

									#!/usr/bin/env python3

									# -*- coding: utf-8 -*-

									import glob

									from os import path

									import os

									import pytesseract

									from PIL import Image

									from queue import Queue

									import threading

									import datetime

									import cv2

									def convertimg(picfile, outdir):

									  '''调整图片大小，对于过大的图片进行压缩

									  picfile:  图片路径

									  outdir：  图片输出路径

									  '''

									  img = Image.open(picfile)

									  width, height = img.size

									  while (width * height > 4000000): # 该数值压缩后的图片大约 两百多k

									    width = width // 2

									    height = height // 2

									  new_img = img.resize((width, height), Image.BILINEAR)

									  new_img.save(path.join(outdir, os.path.basename(picfile)))

									def baiduOCR(ts_queue):

									  while not ts_queue.empty():

									    picfile = ts_queue.get()

									    filename = path.basename(picfile)

									    outfile = 'D:\Study\pythonProject\scrapy\IpProxy\port_zidian.txt'

									    img = cv2.imread(picfile, cv2.IMREAD_COLOR)

									    print("正在识别图片：\t" + filename)

									    message = pytesseract.image_to_string(img,lang = 'eng')

									    message = message.replace('', '')

									    message = message.replace('\n', '')

									    # message = client.basicAccurate(img)  # 通用文字高精度识别，每天 800 次免费

									    #print("识别成功！"))

									    try:

									      filename1 = filename.split('.')[0]

									      filename1 = ''.join(filename1)

									      with open(outfile, 'a+') as fo:

									        fo.writelines('\'' + filename1 + '\'' + ':' + message + ',')

									        fo.writelines('\n')

									        # fo.writelines("+" * 60 + '\n')

									        # fo.writelines("识别图片：\t" + filename + "\n" * 2)

									        # fo.writelines("文本内容：\n")

									        # # 输出文本内容

									        # for text in message.get('words_result'):

									        #   fo.writelines(text.get('words') + '\n')

									        # fo.writelines('\n' * 2)

									      os.remove(filename)

									      print("识别成功！")

									    except:

									      print('识别失败')

									    print("文本导出成功！")

									    print()

									def duqu_tupian(dir):

									  ts_queue = Queue(10000)

									  outdir = dir

									  # if path.exists(outfile):

									  #   os.remove(outfile)

									  if not path.exists(outdir):

									    os.mkdir(outdir)

									  print("压缩过大的图片...")

									  # 首先对过大的图片进行压缩，以提高识别速度，将压缩的图片保存与临时文件夹中

									  try:

									    for picfile in glob.glob(r"D:\Study\pythonProject\scrapy\IpProxy\tmp\*"):

									      convertimg(picfile, outdir)

									    print("图片识别...")

									    for picfile in glob.glob("tmp1/*"):

									      ts_queue.put(picfile)

									      #baiduOCR(picfile, outfile)

									      #os.remove(picfile)

									    print('图片文本提取结束！文本输出结果位于文件中。' )

									    #os.removedirs(outdir)

									    return ts_queue

									  except:

									    print('失败')

									if __name__ == "__main__":

									  start = datetime.datetime.now().replace(microsecond=0)

									  t = 'tmp1'

									  s = duqu_tupian(t)

									  threads = []

									  try:

									    for i in range(100):

									      t = threading.Thread(target=baiduOCR, name='th-' + str(i), kwargs={'ts_queue': s})

									      threads.append(t)

									    for t in threads:

									      t.start()

									    for t in threads:

									      t.join()

									    end = datetime.datetime.now().replace(microsecond=0)

									    print('删除耗时：' + str(end - start))

									  except:

									    print('识别失败')