Vital/IDCardRecogntion/SourceCode/Algorithm/csharp/IDCardAnalyser.cs

using AI.Common.Interface;
using IDCardRecognitionLibs.IDCardRecognitionCore;
using IDCardRecognitionLibs.InferNet;
using IDCardRecognitionLibs.OCRProcessCore;
using IDCardRecognitionLibs.OCRProcessCore.PaddleOCR;
using System.Net;
using System.Text.RegularExpressions;
using System.Xml.Linq;

namespace IDCardRecognitionLibs
{
    public class IDCardAnalyser : IDisposable
    {
        #region public
        private string _netDir = "";
        private IOCRProcessCore? _ocrProcessCore;
        private bool _name = false; // 姓名
        private bool _idNumber = false; // 身份证号码
        private bool _nation = false; //民族
        private bool _address = false; // 地址
        #endregion

        #region public

        /// <summary>
        /// 推理单张图
        /// </summary>
        /// <param name="image"></param>
        /// <returns></returns>
        public IDCardRecogResult EvaluateOneImage(RawImage image)
        {
            try
            {
                // ocr处理（检测、文字提取）
                OCRResult ocrResult = _ocrProcessCore.EvaluateOneImage(image);
                // 整理结果，返回给用户必要的信息
                IDCardRecogResult recogResult = ProcessIdCardInformation(ocrResult);
                return recogResult;
            }
            catch (Exception excep)
            {
                throw new Exception(excep.Message);
            }
        }

        #endregion

        #region constructor

        /// <summary>
        /// 构造函数
        /// </summary>
        /// <param name="dir">模型地址</param>
        /// <param name="config">模型配置项</param>
        /// <param name="parameter">参数</param>
        public IDCardAnalyser(string dir, EnumOCRModel modelType)
        {
            // 根据模型类型选择要实例化哪个类
            _netDir = dir;
            if (modelType == EnumOCRModel.PaddleOCR)
            {
                // 实例化PaddlerOcr
                _ocrProcessCore = new PaddleOCRProcessCore();
                _ocrProcessCore?.LoadModel(_netDir);
            }
            else
            {
                throw new Exception("Other models are not supported yet!");
            }
        }

        #endregion

        public void Dispose()
        {
            if (_ocrProcessCore!= null)
            {
                _ocrProcessCore?.Dispose();
                _ocrProcessCore = null;
            }
        }

        private IDCardRecogResult ProcessIdCardInformation(OCRResult ocrResult)
        {
            IDCardRecogResult idInfor= new IDCardRecogResult();
            idInfor.TimeSpan = ocrResult.TimeAllProcessOneImage;
            string inputTextIn = string.Empty;

            List<string> inputTextList = new List<string> { };
            foreach (var item in ocrResult.TextBlocks)
            {
                inputTextList.Add(item.Text);
                inputTextIn+= item.Text;
            }

            string pattern = "[\\p{P}]";
            string textInforAll = Regex.Replace(inputTextIn, pattern, "");
            // 判断身份证号码是否存在
            string idNumber = ExtractIdNumber(textInforAll);  // 提取身份证号码
            if (string.IsNullOrEmpty(idNumber))
            {
                LogHelper.ErrorLog("IdNumber extraction failed:", string.Empty);
                idInfor.NumerStatus = -1;
                return idInfor;
            }

            _idNumber = true;

            LogHelper.InfoLog("IdNumber extraction successed:" + idNumber);

            idInfor.IdNumber = idNumber;
            textInforAll = textInforAll.Replace(idNumber, "");
            string birthdate, gender;
            ExtractBirthdateAndGender(idNumber, out birthdate, out gender);
            idInfor.Gender = gender;
            idInfor.Birthdate = birthdate;
            LogHelper.InfoLog("Gender extraction successed:" + gender);
            LogHelper.InfoLog("Birthdate extraction successed:" + birthdate);

            List<string> inputTextListIn = new List<string>(inputTextList);
            //inputTextListIn.RemoveAll(text => text.Contains("性别") || text.Contains(idInfor.Gender));
            inputTextListIn.RemoveAll(text => text.Contains("出生") || text.Contains(idInfor.Birthdate));
            inputTextListIn.RemoveAll(text => text.Contains(idNumber));

            string address = string.Empty;
            string name = string.Empty;
            string nation = string.Empty;
            string[] keywordsAll = { "姓名", "民族", "住址" };
            string[] keywordsName = { "姓名" };
            string[] keywordsNation = { "民族" };
            string[] keywordsAddress = { "住址" };


            if (keywordsName.Any(keyword => inputTextIn.Contains(keyword)))
            {
                foreach (string text in inputTextListIn)
                {
                    if (text.Contains("姓名"))
                    {
                        name = ExtractName(text, "姓名");
                        if (!string.IsNullOrEmpty(name))
                        {
                            idInfor.Name = name;
                            textInforAll = textInforAll.Replace("姓名", "").Replace(idInfor.Name, "");
                            inputTextListIn.RemoveAll(text => text.Contains("姓名") || text.Contains(idInfor.Name));
                            _name = true;
                            LogHelper.InfoLog("Name extraction successful:" + name);
                            break;
                        }
                        else
                        {
                            LogHelper.ErrorLog("Name extraction failed:", string.Empty);
                            _name = false;
                            break;
                        }
                    }
                }
            }
            else
            {
                LogHelper.ErrorLog("Name extraction failed:", string.Empty);
                _name = false;
            }


            // 民族
            if (keywordsNation.Any(keyword => inputTextIn.Contains(keyword)))
            {
                foreach (string text in inputTextListIn)
                {
                    if (text.Contains("民族"))
                    {
                        nation = ExtractName(text, "民族");
                        if (!string.IsNullOrEmpty(nation))
                        {
                            idInfor.Nation = nation;
                            textInforAll = textInforAll.Replace("民族", "").Replace(idInfor.Nation, "");
                            inputTextListIn.RemoveAll(text => text.Contains("民族") || text.Contains(idInfor.Nation));
                            _nation = true;
                            LogHelper.InfoLog("Nation extraction successful:" + nation);
                            break;
                        }
                        else
                        {
                            LogHelper.ErrorLog("Nation extraction failed:", string.Empty);
                            _nation = false;
                            break;
                        }
                    }
                }
            }
            else
            {
                LogHelper.ErrorLog("Nation extraction failed:", string.Empty);
                _nation = false;
            }

            //inputTextListIn.RemoveAll(text => text.Contains("性别") || text.Contains(idInfor.Gender));

            // 住址
            if (keywordsAddress.Any(keyword => inputTextIn.Contains(keyword)))
            {
                foreach (string text in inputTextListIn)
                {
                    if (text.Contains("住址"))
                    {
                        if (!textInforAll.Contains("公民身份号码"))
                        {
                            textInforAll = inputTextIn.Replace(idNumber, "公民身份号码");
                        }
                        var patternAdd = @"住址([\s\S]*?)公民身份号码";
                        Match match = Regex.Match(textInforAll, patternAdd);
                        address = match.Groups[1].Value.Trim().Replace(" ", "");
                        // 判断地址是否全
                        if (string.IsNullOrEmpty(address))
                        {
                            _address = false;
                            LogHelper.ErrorLog("Address extraction failed:", string.Empty);
                        }
                        else if (address.Contains("省") || address.Contains("北京市") || address.Contains("上海市") || address.Contains("天津市") || address.Contains("自治区") || address.Contains("重庆市"))
                        {
                            idInfor.Address = address;
                            _address = true;
                            LogHelper.InfoLog("Address extraction successful:" + address);
                            break;
                        }
                        else
                        {
                            inputTextListIn.Add(address);
                            foreach (string textIn in inputTextListIn)
                            {
                                if (text.Contains("省"))
                                {
                                    idInfor.Address = text + address;
                                    _address = true;
                                    LogHelper.InfoLog("Address extraction successful:" + address);
                                    break;
                                }
                            }
                        }
                    }
                }
            }
            else
            {
                LogHelper.ErrorLog("Address extraction failed:", string.Empty);
            }


            // 判断检测结果有哪些
            if (_name && _nation && _address)
            {
                idInfor.NumerStatus = 1;
                LogHelper.InfoLog("IDCardInfor extraction successful,NumerStatus: 1");
            }
            else if (_name || _nation || _address)
            {
                idInfor.NumerStatus = 2;
                LogHelper.InfoLog("Part IDCardInfor extraction successful, NumerStatus: 2");
            }
            else if (!_name && !_nation && !_address)
            {
                idInfor.NumerStatus = 3;
                LogHelper.InfoLog("Part IDCardInfor extraction successful,NumerStatus: 3");
            }

            return idInfor;

        }

        /// <summary>
        /// 
        /// </summary>
        /// <param name="input"></param>
        /// <returns></returns>
        private static string ExtractName(string input, string searchKeyword)
        {
            // 从输入字符串中提取地址的正则表达式模式，包含数字
            string pattern = $"{searchKeyword}([\u4e00-\u9fa5\\d]+)";
            Regex regex = new Regex(pattern);
            Match match = regex.Match(input);

            // 如果匹配成功，返回匹配的地址信息（包含数字）；否则返回空字符串
            return match.Success ? match.Groups[1].Value : string.Empty;
        }


        /// <summary>
        /// 提取身份证号码
        /// </summary>
        /// <param name="input"></param>
        /// <returns></returns>
        private static string ExtractIdNumber(string input)
        {
            // 从输入字符串中提取身份证号码
            string pattern = @"(\d{17}[\dxX])[\dxX]?";
            Regex regex = new Regex(pattern);
            Match match = regex.Match(input);
            return match.Success ? match.Value : string.Empty;
        }


        /// <summary>
        /// 根据身份证号码提取性别 出生
        /// </summary>
        /// <param name="idNumber"></param>
        /// <returns></returns>
        private static void ExtractBirthdateAndGender(string idNumber, out string birthdate, out string gender)
        {
            birthdate = string.Empty;
            gender = string.Empty;

            if (idNumber.Length == 18)
            {
                // 提取前17位作为日期和顺序码
                string dateAndSeq = idNumber.Substring(6, 8);

                // 提取年、月、日
                int year = int.Parse(dateAndSeq.Substring(0, 4));
                int month = int.Parse(dateAndSeq.Substring(4, 2));
                int day = int.Parse(dateAndSeq.Substring(6, 2));

                // 构造日期字符串
                birthdate = $"{year}年{month}月{day}日";

                // 提取性别信息
                int genderCode = int.Parse(idNumber.Substring(16, 1));
                gender = genderCode % 2 == 0 ? "女" : "男";
            }
        }

    }
}