Introduction
I promised the expert Zhang Shanyou from the park to write a series of blogs on AI, so I started my AI series journey.
1. Introduction
The API documentation for ID card recognition is available at: http://ai.baidu.com/docs#/OCR-API/top
Interface Description
The user requests the service to recognize the ID card, which includes both the front and back.
Request Instructions
Request Example
HTTP Method: <span>POST</span>
Request URL: https://aip.baidubce.com/rest/2.0/ocr/v1/idcard
Note: You need to become a Baidu developer to obtain the API key and Secret Key.
Obtaining Access_Token
The Baidu Access_token has a time limit, approximately 30 days, so it is recommended to encapsulate a functional method to call the latest one each time.
-
access_token: The Access Token to be obtained;
-
expires_in: The validity period of the Access Token (in seconds, generally one month);
2. Technical Implementation
Baidu’s text recognition provides SDKs. If there are supported languages, you can use the SDK directly. The author used an HTTP request encapsulation.
For images, there are size requirements; image data must be base64 encoded and then URL encoded. The size after base64 encoding and URL encoding should not exceed 4M, the shortest side must be at least 15px, the longest side must be at most 4096px, and formats supported are jpg/png/bmp.
Basic Interface Encapsulation
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>
/// Get access token, normal JSON entity model returned by Baidu interface
/// </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>
/// Get access token, failed JSON entity model returned by Baidu interface
/// </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>
/// Text recognition – ID card recognition application (only retrieves ID card image information, does not connect to the Ministry of Public Security, cannot verify authenticity, only recognizes text from the image)
/// </summary>
public class IDCardRecognition
{
// ID card recognition
/// <summary>
/// ID card recognition
/// </summary>
/// <param name=”token”>Access token</param>
/// <param name=”imagePath”>Image path</param>
/// <param name=”recognitionString”>Recognition result</param>
/// <param name=”errorMsg”>Error message</param>
/// <param name=”id_card_side”> front: ID card front; back: ID card back</param>
/// <param name=”detect_direction”>Whether to detect image orientation, defaults to not detecting, i.e.,: false. Orientation refers to whether the input image is in the normal direction, rotated 90/180/270 degrees counterclockwise. Optional values include: – true: detect orientation; – false: do not detect orientation.</param>
/// <param name=”detect_risk”> string type whether to enable ID card risk type (ID card copy, temporary ID card, re-photographed ID card, modified ID card) functionality, defaults to not enabling, i.e.,: false. Optional values: true – enable; false – not enable</param>
/// <returns>Result status</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 Basic Verification
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); // Base64 encoding of the image
Encoding encoding = Encoding.Default;
string urlEncodeImage = HttpUtility.UrlEncode(strbaser64);
byte[] tempBuffer = encoding.GetBytes(urlEncodeImage);
if (tempBuffer.Length > 1024 * 1024 * 4)
{
errorMsg += “Image size after encryption exceeds 4MB!”;
recognitionString = “”;
tempModel.state = false;
tempModel.errorMsg = errorMsg;
return tempModel;
}
#endregion
#region Request Interface
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\””))// indicates an exception
{
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)// External exception, such as network exception
{
resultState = false;
errorMsg = ex.ToString();
tempModel.state = false;
tempModel.errorMsg = ex.ToString();
return tempModel;
}
}
}
}
Core part of Winform invocation
/// <summary>
/// Recognition operation
/// </summary>
/// <param name=”filePath”></param>
/// <param name=”id_card_side”>Front or back of the ID card</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();// Main thread executes progress bar, sub-thread performs data request operation
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 —————————Recognition Successful——————————-“);
tb_showInfo.AppendText(“\r\n” + result.successModel.ToJson() + “\r\n”);
}));
}
else
{
this.Invoke(new Action(() =>
tb_showInfo.AppendText(“\r\n—————————–Recognition Failed!——————————–“);
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();
}
Effect as shown: The ID card in the image is searched from Baidu Tieba, authenticity unknown.
PS: This is only text recognition and does not connect to the Ministry of Public Security for verification (identity validity recognition). To connect to the Ministry of Public Security for recognition, a fee is required.
3. Integrated Application
The author’s application is written in a plugin-based hot-swappable mode, encapsulating each interface as a plugin, dynamically combined using injection.
To facilitate a user-friendly experience, a progress bar is included in the request, using a new thread for interface requests to prevent the interface from freezing.
Related Articles:
-
Build 2017 | A Comprehensive Overview of Microsoft Build 2017 Conference: Bringing AI to the Edge
-
Understanding Microsoft Visual Studio Tools for AI
-
VS Tools for AI Complete Guide
-
VS Tools for AI Complete Guide (2) Low-Configuration Virtual Machines Can Also Handle Deep Learning Without NC/NV Series
-
Based on Emgu CV + Baidu Face Recognition, Implementing Video Dynamic Face Capture and Recognition
Original text: http://www.cnblogs.com/linbin524/p/BaiduOCR_IDCard.html
.NET Community News, In-depth Articles, Welcome to Visit the Public Account Article Summary http://www.csharpkit.com