Python识别图片中的文字

1 import os,glob
2 def photo_compression(original_imgage,tmp_image_path):
3 '''图片备份、压缩；param original_imgage:原始图片路径；param tmp_imgage_path:临时图片路径，备份路径；return'''
4 from PIL import Image
5 img = Image.open(original_imgage)
6 width,height = img.size
7 while (width*height>4000000):#该数值压缩后的图片大约200多k
8 width = width//2
9 height = height//2
10 e_img = img.resize((width,height),Image.BILINEAR)
11 save_path = os.path.join(tmp_image_path,os.path.basename(original_imgage))
12 e_img.save(save_path)
13 return save_path
14
15 def ocr(original_image):
16 '''使用百度OCR进行文字识别，支持JPG、JPEG、PNG、BMP格式；param original_image:待识别图片；return'''
17 from aip import AipOcr
18 filename = os.path.basename(original_image)
19 #输入自己的百度ai账号ID密码：参考链接：https://m.toutiaocdn.com/i6704242394566492684/
20 APP_ID = '******'
21 API_KEY = '*******'
22 SECRECT_KEY = '*********'
23
24 client = AipOcr(APP_ID,API_KEY,SECRECT_KEY)
25
26 with open(original_image,'rb') as picfile_read:
27 img = picfile_read.read()
28 print('正在识别图片：{0}......'.format(filename))
29 try:
30 result = client.basicGeneral(img)#通用文字识别，50000次/天免费
31 except:
32 result = client.basicAccurate(img)#通用文字识别（高精度版），500次/天免费
33 return result
34
35 def run_ocr(original_image,tmp_image_path,result_file_path='identify_results.txt'):
36 '''主函数 批量执行图片文本识别，结果存储；original_image:原始图片；tmp_image_path:临时图片；result_file_path:识别文字存储文件；return'''
37
38 if os.path.exists(result_file_path):#判断是否存在历史识别结果，若存在则删除
39 os.remove(result_file_path)
40 if not os.path.exists(tmp_image_path):#判断临时图片路径是否存在，若不存在则创建
41 os.mkdir(tmp_image_path)
42 tmp_file_path = []#临时文件路径列表
43 for picfile in glob.glob(original_image):#glob.glob的参数是一个只含有方括号、问号、正斜线的正则表达式
44 tmp_file = photo_compression(picfile,tmp_image_path)
45 tmp_file_path.append(tmp_file)
46 for picfile in tmp_file_path:#遍历所有文件，进行OCR识别 结果存储
47 result = ocr(picfile)
48 lines = [text.get('words') + '\n' for text in result.get('words_result')]
49 # lines = [text.get('words').encode('utf-8')+'\n' for text in result.get('words_result')]
50
51 with open(result_file_path,'a+',encoding='utf-8') as fo:
52 fo.writelines("="*100+'\n')
53 fo.writelines("【识别图片】：{0} \n".format(os.path.basename(picfile)))
54 fo.writelines("【文本内容】： \n")
55 fo.writelines(lines)
56 os.remove(picfile)
57
58 if __name__ == '__main__':
59 tmp_image_path = os.getcwd()+'\\tmp'
60 original_image = os.getcwd() + '\\*.png'
61 run_ocr(original_image,tmp_image_path)