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)