Python中AI圖像識別實現身份證識別
圖像識別說白了就是把一張照片上面的文字進行提取，提供工作效率
需求分析身份證識別主要是把一張身份證照片上面的文字信息進行提取，不用再使用人工去手動抄寫了，下面給大家說的這個身份識別主要是使用python+flask+華為云OCR進行實現的。
步驟 申請華為云OCR接口 獲取token 調用身份證識別接口 提取身份證信息申請華為云OCR接口圖像識別主要使用的就是華為云OCR平臺申請的接口，申請地址為：“https://www.huaweicloud.com”。訪問申請的地址后點擊菜單欄中的“控制臺”，點擊后會進入登錄頁面。
登錄成功后進入控制臺，依次單擊“服務列表”-“人工智能”-“文字識別OCR”
進入后點擊身份證識別后面的“開通服務”即可，（這里需要主要的是免費開通的身份證識別服務僅有1000次的免費調用次數，超過次數將會收費，建議購買套餐包，購買可以找我，我可以給優惠哦）
獲取token值時你需要知道你的以下信息：
IAM用戶所屬賬號名 IAM用戶名 IAM用戶的登錄密碼 IAM用戶所屬賬號的項目名稱首先就是獲取IAM用戶所屬賬號名，IAM用戶所屬賬號名為你登陸的那個賬號
獲取IAM用戶名，選擇右上角“用戶名”-點擊“統一身份認證”
里面的用戶即為IAM用戶名
獲取IAM用戶所屬賬號的項目名稱，點擊“項目”
里面的項目很多，你只需要記錄你開啟OCR身份證識別區域的項目名稱ID就可以了，比如我選擇的北京四所以我的項目名稱就是：“cn-north-4”
知道了這些信息之后，后面就是如何通過這些信息去獲取token了，下面我就直接上代碼了哦
import requests# 下面url為華為云獲取token的接口，這個接口基本一致，需要注意的是“iam.cn-north-4.myhuaweicloud.com”這個需要根據具體需求進行修改，不同區域項目不同url = 'https://iam.cn-north-4.myhuaweicloud.com/v3/auth/tokens'# 頭部信息headers = {’Content-Type’: ’application/json’}# json內容，需要上傳進行做驗證payload = { 'auth': {'identity': { 'methods': ['password' ], 'password': {'user': {# name值為你的IAM賬號名 'name': 'XXX', # password值為你的IAM密碼 'password': 'XXX', 'domain': { # name值為你的主賬號的賬號名'name': 'XXX' }} }},'scope': { 'project': { # name值為你的項目ID'name': 'cn-north-4' }} }}# 使用POST上傳頭部內容和body內容response = requests.post(url, headers=headers, json=payload)# 獲取token值token = response.headers[’X-Subject-Token’]# 打印token值print(token)
打印出來的值是類似于下面這種的字符串
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-1238KehFzgXhJe59fOYyc7hz+qZ5456iYD0pSGqjICZepOBeTvywtsE2fgHdlQq4RZMDm1bkVj46OVNGszzpz8IGecZxRDhxDfBbVsllo9T41QD28o9QZUNyhhXa-ZIS2UdDK-q7gBsR3mT2LRqjQ+J6vpcCzAe6v71IHAf77lgE+vwg7f6qOfhYjRNB9kSC2L1YxJ1eS9j5pzuqlnK+9BOv2U3BIaKsP9yNiP-luS78LzwuEG2rD9OInMst7bCr7HeWT2bZ8gCT4567BvsONvGqVvr3AA==
調用身份證識別接口下面就是調用識別接口，調用這個接口需要了解以下信息
URL Header參數 Body參數首先是URLPOST https://{endpoint}/v2/{project_id}/ocr/id-card
參數 說明 endpoint 項目接口域名，這里還是需要確定開通服務的區域，每個區域不同的域名 project_id 項目ID，不出意外的話就是你上面獲取token的項目ID
請求參數請求Header參數
參數 說明 X-Auth-Token 用戶Token，就是上面獲取到的token值 Content-Type 發送的實體的MIME類型，參數值為“application/json”
請求Body參數
參數 說明 image 圖像數據，base64編碼，要求base64編碼后大小不超過10MB。圖片最小邊不小于15px，最長邊不超過8000px。支持JPEG、JPG、PNG、BMP、TIFF格式。 url 圖片的url路徑 side front：身份證人像面，back：身份證國徽面
了解了上面的一系列信息之后，后面就是如何寫代碼了，下面就是直接代碼
import requestsimport base64# url值url = https://{endpoint}/v2/{project_id}/ocr/id-card# 上面獲取到的token值token = ''# 設置頭部信息headers = {’Content-Type’: ’application/json’, ’X-Auth-Token’: token}# 圖片路徑imagepath = r’./static/XXX.jpg’# 設置圖片with open(imagepath, 'rb') as bin_data: image_data = bin_data.read()# 使用圖片的base64編碼image_base64 = base64.b64encode(image_data).decode('utf-8')# url與image參數二選一payload = {'image': image_base64}# 上傳參數response = requests.post(url, headers=headers, json=payload)# 打印識別的信息response = response.textprint(response)
最后識別出來的結果類似于下面的信息：
{ 'result': { 'name': 'XXX', 'sex': '男', 'ethnicity': '漢', 'birth': 'XXX', 'address': 'XXX', 'number': 'XXX' }}
如果想獲取其中的某一個值的話需要先轉換為字典，通過字典再進行單一讀取
import json# 轉換為字典test = json.loads(response)# 從字典中獲取相應的值username = test[’result’][’name’]
到這里就已經完成了這個項目
總結上面已經簡單說明了如何調用相應的接口去識別身份證了，下面我把源碼給到大家吧，大家可以作為參考哦，下面是使用python+flask+redis+華為云OCR實現的
import jsonfrom flask import Flask, render_templateimport requestsfrom flask_redis import FlaskRedisimport base64app = Flask(__name__)# 設置redis連接地址REDIS_URL = ’redis://127.0.0.1:6379/0’# 實例化redisrd = FlaskRedis(app, decode_responses=True)@app.route(’/’)def hello_world():# 獲取redis緩存中token token = rd.get(’token’) # 判斷redis中是否有token值，有則直接使用，沒有則獲取token值并寫入redis中 if token is None:url = 'https://iam.cn-north-4.myhuaweicloud.com/v3/auth/tokens'headers = {’Content-Type’: ’application/json’}payload = { 'auth': {'identity': { 'methods': ['password' ], 'password': {'user': { 'name': 'XXX', 'password': 'XXX', 'domain': {'name': 'XXX' }} }},'scope': { 'project': {'name': 'cn-north-4' }} }}response = requests.post(url, headers=headers, json=payload)requst = response.ok# 判斷是否獲取token值成功if not requst: error = '獲取token失敗' return render_template(’test.html’, db=error)# 從返回值中獲取token值 token = response.headers[’X-Subject-Token’]# 在redis添加token值，下次不再調用，并設置緩存時間為24小時relue = rd.set(’token’, token, ex=60 * 1440)# 判斷是否再redis中添加token值成功if relue != 1: error = '寫入錯誤' return render_template(’test.html’, db=error) url = 'https://ocr.cn-north-4.myhuaweicloud.com/v2/cn-north-4/ocr/id-card' headers = {’Content-Type’: ’application/json’, ’X-Auth-Token’: token} imagepath = r’./static/XXX.jpg’ with open(imagepath, 'rb') as bin_data:image_data = bin_data.read() image_base64 = base64.b64encode(image_data).decode('utf-8') # 使用圖片的base64編碼 payload = {'image': image_base64} # url與image參數二選一 response = requests.post(url, headers=headers, json=payload) # 判斷是否識別成功 if not response.ok:error = '識別失敗，請聯系管理員'return render_template(’test.html’, db=error) response = response.text test = json.loads(response) username = test[’result’][’name’] return render_template(’test.html’, db=username)if __name__ == ’__main__’: app.run()
到此這篇關于Python中AI圖像識別實現身份證識別的文章就介紹到這了,更多相關Python AI圖像識別 內容請搜索好吧啦網以前的文章或繼續瀏覽下面的相關文章希望大家以后多多支持好吧啦網！
相關文章：
網公網安備