一、主要步骤

2. 它让我们先获取 access_token，先看获取access_token的文档

https://ai.baidu.com/docs#/Auth/top

3. 需要 app_key secrect_key 需要先自行申请好

二、开始coding

1. 先定义一个baidu_token类用来处理token相关的工作

class baidu_token:

__app_key = '申请到的app_key'

__secrect_key = '申请到的secrect_key'

def __init__(self):

self._access_token_url = 'https://aip.baidubce.com/oauth/2.0/token?grant_type=client_credentials&client_id=%s&client_secret=%s'

self._access_token_url = self._access_token_url % (self.__app_key, self.__secrect_key)

self._ok = False

self._token_js = None

self._get_token()

def _get_token(self, cb=print):

try:

rsp = requests.get(self._access_token_url)

if rsp.status_code == 200:

self._token_js = json.loads(rsp.content.decode())

if self._token_js.get('access_token'):

self._ok = True

except:

pass

def access_token(self):

if self._ok:

return self._token_js['access_token']

2. 定义一个百度识别身份证照片的类

class baidu_card_img:

__request_url = "https://aip.baidubce.com/rest/2.0/ocr/v1/general_basic"

def __init__(self, baidu_token = None, img_path = None):

self._img_path = img_path

self._token = baidu_token

def _get_data(self):

if not self._token:

print('access_token invalid')

return

access_token = self._token.access_token()

if not access_token:

print('access_token invalid')

return

try:

#二进制读取图片 使用base64编码

with open(self._img_path, 'rb') as f:

data = base64.b64encode(f.read())

except:

print('img_path invalid')

return

#组装baidu API需要的参数

param = {"image":data}

url = self.__request_url + '?access_token=' + access_token

#headers

headers = {"Content-Type":"application/x-www-form-urlencoded"}

rsp = requests.post(url, data=param, headers=headers)

if rsp.status_code == 200:

return json.loads(rsp.content.decode())

def get_image_data(self, token = None, image_path = None):

self._img_path = image_path or self._img_path

self._token = token or self._token

ret_js = self._get_data()

if not ret_js:

return

lst_words = ret_js.get('words_result')

if not lst_words or len(lst_words) < 5:

print('api return nothing')

return

real_words = []

for i in lst_words:

#测试发现 身份证上面那行英文字母会被识别出来 用正则过滤掉字母开头的结果

#被str.isalpha()坑了 函数返回的是字符串全英文或中文 不能用来过滤纯英文

if not re.match(r'[a-zA-Z]', i['words'][0]):

real_words.append(i)

lst_words = real_words

result = {}

result['name'] = lst_words[0]['words'][2:]

result['sex'] = lst_words[1]['words'][2]

result['nation'] = lst_words[1]['words'][5:]

result['born'] = lst_words[2]['words'][2:]

result['address'] = lst_words[3]['words'][2:]

#测试发现 如果身份证地址信息可能占一行 也可能占两行

if len(lst_words) > 5:

result['address'] += lst_words[4]['words']

result['card_num'] = lst_words[len(lst_words) - 1]['words'][6:]

else:

result['card_num'] = lst_words[4]['words'][6:]

return result

写个demo测试一下结果

card = baidu_card_img(baidu_token())

dir_path = './test_img/'

dirs = os.listdir(dir_path)

for file in dirs:

if len(file) < 4:

continue

suffix = file[-4:]

if suffix == '.jpg' or suffix == '.png':

abs_path = dir_path + file

data = card.get_image_data(image_path=abs_path)

print(data)

结果如图:(图片是网图 我自己也拿身份拍了一张试了也ok)