百度OCR文字识别-身份证识别

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

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

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

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");

}));

}