百度OCR文字识别-身份证识别

亦凉 2022-03-27 10:58 443阅读 0赞

**简介**

答应了园区大牛张善友 要写AI 的系列博客，所以开始了AI 系列之旅。

**一、介绍**

身份证识别 API 接口文档地址：http://ai.baidu.com/docs\#/OCR-API/top

## 接口描述 ##

用户向服务请求识别身份证，身份证识别包括正面和背面。

## 请求说明 ##

请求示例

HTTP 方法：`POST`

请求URL： `https://aip.baidubce.com/rest/2.0/ocr/v1/idcard`

备注：你需要 成为百度开发者，获取API key 和Secret Key

![640?wx\_fmt=png&wxfrom=5&wx\_lazy=1][640_wx_fmt_png_wxfrom_5_wx_lazy_1]

Access\_Token 的获取

百度Access\_token 有效期有时间限制，大概是30天左右，所以建议封装成功能方法每次调用最新的。

*  access\_token：要获取的Access Token；
 *  expires\_in：Access Token的有效期(秒为单位，一般为1个月)；

**二、技术实现**

百度 文字识别 有提供SDK。如果有支持的语言，可以直接用sdk。笔者自己用的Http 请求封装

![640?wx\_fmt=png&wxfrom=5&wx\_lazy=1][640_wx_fmt_png_wxfrom_5_wx_lazy_1 1]

对于图片大小有要求的，图像数据，base64编码后进行urlencode，要求base64编码和urlencode后大小不超过4M，最短边至少15px，最长边最大4096px,支持jpg/png/bmp格式

![0?wx\_fmt=png][0_wx_fmt_png]

接口基础封装

using System;

using System.Collections.Generic;

using System.Linq;

using System.Text;

using System.Threading.Tasks;

namespace BaiduAIAPI.Model

public class AccessTokenModel \{

public bool IsSuccess \{ get; set; \}

public SuccessAccessTokenModel SuccessModel \{ get; set; \}

public ErrorAccessTokenModel ErrorModel \{ get; set; \}

/// <summary>

/// 获取accesstoken，正常 的 百度接口返回的json 实体模型

/// </summary>

public class SuccessAccessTokenModel

public string refresh\_token \{ get; set; \}

public int expires\_in \{ get; set; \}

public string scope \{ get; set; \}

public string session\_key \{ get; set; \}

public string session\_secret \{ get; set; \}

public string access\_token \{ get; set; \}

/// <summary>

/// 获取accesstoken，失败的 百度接口返回的json 实体模型

/// </summary>

public class ErrorAccessTokenModel

public string error \{ get; set; \}

public string error\_description \{ get; set; \}

using System;

using System.IO;

using System.Net;

using System.Text;

using System.Web;

using AOP.Common;

using AOP.Common.DataConversion;

using BaiduAIAPI.Model;

using BaiduAIAPI.Type;

namespace BaiduAIAPI.ORC\_Characterbase64

/// <summary>

/// 文字识别--身份证识别 应用（只是获取身份证图片 信息，没有和公安部联网，无法确认真假，只是单纯从图片上识别文字）

/// </summary>

public class IDCardRecognition

// 身份证识别

/// <summary>

/// 身份证识别

/// </summary>

/// <param name="token">Accesstoken</param>

/// <param name="imagePath">图片路径</param>

/// <param name="recognitionString">识别结果</param>

/// <param name="errorMsg">错误信息</param>

/// <param name="id\_card\_side"> front：身份证正面；back：身份证背面</param>

/// <param name="detect\_direction">是否检测图像朝向，默认不检测，即：false。朝向是指输入图像是正常方向、逆时针旋转90/180/270度。可选值包括:- true：检测朝向；- false：不检测朝向。</param>

/// <param name="detect\_risk"> string 类型 是否开启身份证风险类型(身份证复印件、临时身份证、身份证翻拍、修改过的身份证)功能，默认不开启，即：false。可选值:true-开启；false-不开启</param>

/// <returns>结果状态</returns>

public static IDCardRecognitionModel GetIdcardRecognitionString(string token, string imagePath, ref string recognitionString, out string errorMsg, string id\_card\_side="front", bool detect\_direction=false, string detect\_risk="false")

bool resultState = true;

IDCardRecognitionModel tempModel = new IDCardRecognitionModel();

try

\#region 基础校验

string verificationMsg = "";

errorMsg = "";

bool isVerification = ImageVerification.VerificationImage(imagePath, out verificationMsg);

if (!isVerification)

errorMsg += verificationMsg;

tempModel.state = false;

tempModel.errorMsg = errorMsg;

return tempModel;

string strbaser64 = ConvertDataFormatAndImage.ImageToByte64String(imagePath, System.Drawing.Imaging.ImageFormat.Jpeg); // 图片的base64编码

Encoding encoding = Encoding.Default;

string urlEncodeImage = HttpUtility.UrlEncode(strbaser64);

byte\[\] tempBuffer = encoding.GetBytes(urlEncodeImage);

if (tempBuffer.Length > 1024 \* 1024 \* 4)

errorMsg += "图片加密 后的大小超过4MB！";

recognitionString = "";

tempModel.state = false;

tempModel.errorMsg = errorMsg;

return tempModel;

\#endregion

\#region 请求接口

recognitionString = "";

string host = "https://aip.baidubce.com/rest/2.0/ocr/v1/idcard?access\_token=" + token;

String str = "id\_card\_side=" + id\_card\_side + "&detect\_direction=" + detect\_direction + "&detect\_risk=" + detect\_risk + "&image=" + HttpUtility.UrlEncode(strbaser64);

var tempResult = HttpRequestHelper.Post(host, str);

recognitionString = tempResult;

if (recognitionString.Contains("\\"error\_code\\""))//说明异常

resultState = false;

tempModel.state = false;

tempModel.errorTypeModel = Json.ToObject<ErrorTypeModel>(tempResult);

tempModel.errorTypeModel.error\_discription = ORC\_CharacterRecognitionErrorType.GetErrorCodeToDescription(tempModel.errorTypeModel.error\_code);

else

tempModel.state = true;

tempModel.successModel = Json.ToObject<IDCardRecognitionSuccessResultModel>(tempResult);

\#endregion

return tempModel;

catch (Exception ex)//接口外部异常，如网络异常

resultState = false;

errorMsg = ex.ToString();

tempModel.state = false;

tempModel.errorMsg = ex.ToString();

return tempModel;

winform 调用核心部分

/// <summary>

/// 识别操作

/// </summary>

/// <param name="filePath"></param>

/// <param name="id\_card\_side">身份证 正面还是背面</param>

/// <param name="detect\_direction"></param>

/// <param name="detect\_risk"></param>

public void Distinguish(string filePath, string id\_card\_side = "front", bool detect\_direction = false, string detect\_risk = "false")

DoTime();//主线程执行进度条，子线程进行数据请求操作

t1 = new Thread(new ThreadStart(() =>

var temp = BaiduAIAPI.Access\_Token.GetAccessToken();

if (temp.IsSuccess)

string data = "";

string error = "";

var result = IDCardRecognition.GetIdcardRecognitionString(temp.SuccessModel.access\_token, filePath, ref data, out error, id\_card\_side, detect\_direction, detect\_risk);

this.Invoke(new Action(() =>

tb\_showInfo.AppendText("\\r\\n -----------------------------------------------------------------");

\}));

if (result.state)

this.Invoke(new Action(() =>

tb\_showInfo.AppendText("\\r\\n ---------------------------识别成功-------------------------------");

tb\_showInfo.AppendText("\\r\\n" + result.successModel.ToJson() + "\\r\\n");

\}));

else

this.Invoke(new Action(() =>

tb\_showInfo.AppendText("\\r\\n-----------------------------识别失败！--------------------------------");

tb\_showInfo.AppendText("\\r\\n" + result.successModel.ToJson() + result.errorMsg + "\\r\\n");

\}));

else

this.Invoke(new Action(() =>

AttrMessage.ErrorMsg(temp.ErrorModel.error);

\}));

this.Invoke(new Action(() =>

progressBar\_ToReadDistinguish.Value = 100;

timer1.Enabled = false;

progressBar\_ToReadDistinguish.Value = 0;

\}));

t1.IsBackground = true;

t1.Start();

效果如图：图中的身份证是我百度贴吧搜索的，不知道真伪。

![0?wx\_fmt=png][0_wx_fmt_png 1]

PS：这个只是文字识别，并不是真正公安部联网识别（身份有效性识别），要连接公安部识别需要 付费。

**三、整合应用**

**笔者的应用是结合自己写的插件化热插拔模式写的，把每个接口封装成为一个插件，采用注入形式动态化结合**

**![0?wx\_fmt=png][0_wx_fmt_png 2]**

为了便于友好用户体验，在请求使用加入进度条，采用新的线程去进行接口请求，防止 界面卡住。

*  [Build 2017 | 一文看懂微软 Build 2017 大会：让 AI 走向边缘][Build 2017 _ _ Build 2017 _ AI]
 *  [认识微软Visual Studio Tools for AI][Visual Studio Tools for AI]
 *  [VS Tools for AI全攻略][VS Tools for AI]
 *  [VS Tools for AI全攻略（2）低配置虚拟机也能玩转深度学习，无需NC/NV系列][VS Tools for AI_2_NC_NV]
 *  [基于Emgu CV+百度人脸识别，实现视频动态 人脸抓取与识别][Emgu CV_]

原文：http://www.cnblogs.com/linbin524/p/BaiduOCR\_IDCard.html

--------------------

**.NET社区新闻，深度好文，欢迎访问公众号文章汇总 http://www.csharpkit.com**

![640?wx\_fmt=jpeg][640_wx_fmt_jpeg]

再分享一下我老师大神的人工智能教程吧。零基础！通俗易懂！风趣幽默！还带黄段子！希望你也加入到我们人工智能的队伍中来！[https://blog.csdn.net/jiangjunshow][https_blog.csdn.net_jiangjunshow]

[640_wx_fmt_png_wxfrom_5_wx_lazy_1]: https://img-blog.csdnimg.cn/img_convert/41cc11b6a8ca97b8e4928b369701b9cb.png;wxfrom=5&wx_lazy=1
[640_wx_fmt_png_wxfrom_5_wx_lazy_1 1]: https://img-blog.csdnimg.cn/img_convert/028ab8f282e0afd4a82c71b2c76e95ec.png;wxfrom=5&wx_lazy=1
[0_wx_fmt_png]: /images/20220327/6458e0df307148d389f7cc0957a569a6.png
[0_wx_fmt_png 1]: /images/20220327/06b0bf051934416c9dfddeab6968dc3e.png
[0_wx_fmt_png 2]: /images/20220327/eef5a0ed33154413992709ee95ff56ef.png
[Build 2017 _ _ Build 2017 _ AI]: http://mp.weixin.qq.com/s?__biz=MzAwNTMxMzg1MA==&mid=2654068695&idx=2&sn=257c440df7df06faea51d0db41434757&chksm=80dbfb82b7ac729494bd76c5dc75436cbf2abef1db3f04d81ebd8757c19afbe996bf7158d79c&scene=21#wechat_redirect
[Visual Studio Tools for AI]: http://mp.weixin.qq.com/s?__biz=MzAwNTMxMzg1MA==&mid=2654069570&idx=3&sn=5fb5b6b66d1d62bdedeca6ac64bdef3e&chksm=80dbc717b7ac4e010a47f601b140c410d7bf388d18cc75016b16c5eb3e0313d0f9df51a2f103&scene=21#wechat_redirect
[VS Tools for AI]: http://mp.weixin.qq.com/s?__biz=MzAwNTMxMzg1MA==&mid=2654069598&idx=2&sn=713bb1d74cb38acbd77c1e77e75f6bf9&chksm=80dbc70bb7ac4e1d870b6b3deb5b35b9d824303c6a467b599456b8fdd05e24e9b21de642cb0f&scene=21#wechat_redirect
[VS Tools for AI_2_NC_NV]: http://mp.weixin.qq.com/s?__biz=MzAwNTMxMzg1MA==&mid=2654069604&idx=3&sn=421937fe0aa7798a14d030a7b1f23749&chksm=80dbc731b7ac4e277aa1bcc6d845adfd36cfdf6e1af5ba2ad084d9ce0803eaad7cbf947606b0&scene=21#wechat_redirect
[Emgu CV_]: http://mp.weixin.qq.com/s?__biz=MzAwNTMxMzg1MA==&mid=2654069619&idx=1&sn=30b11680a33c4ee6a8a42f03da845da5&chksm=80dbc726b7ac4e3080c4c40db753ba0a6d67876ee1d01516e0fc86a6938547d160084a59be9a&scene=21#wechat_redirect
[640_wx_fmt_jpeg]: /images/20220327/66144b06b21e48338582e1d44611e0ca.png
[https_blog.csdn.net_jiangjunshow]: https://blog.csdn.net/jiangjunshow/article/details/77338485