C# paddlerocrsharp识别身份证号-编程知识

C# paddlerocrsharp识别身份证号

https://gitee.com/raoyutian/paddle-ocrsharp

项目搭建

新建控制台项目
安装paddleocrsharp
在这里插入图片描述
下载训练好的模型

解压放到对应的文件夹中，都修改为如果较新则复制

编写代码OCRHelper.cs

using PaddleOCRSharp;namespace OCRTest02;public class OCRHelper
{/// <summary>/// 初始化OCR引擎/// </summary>/// <returns></returns>public static PaddleOCREngine initOcrEngine(string exePath){//OCR参数OCRParameter oCRParameter = new OCRParameter();oCRParameter.cpu_math_library_num_threads = 6;//预测并发线程数,CPU预测时的线程数，在机器核数充足的情况下，该值越大，预测速度越快；默认10oCRParameter.enable_mkldnn = true;//web部署该值建议设置为0,否则出错，内存如果使用很大，建议该值也设置为0.oCRParameter.cls = true; //是否执行文字方向分类；默认falseoCRParameter.use_angle_cls = true;//是否开启方向检测，用于检测识别180旋转oCRParameter.det_db_score_mode = false;//是否使用多段线，即文字区域是用多段线还是用矩形，oCRParameter.det_db_unclip_ratio = 1.6f;//OCR配置OCRModelConfig config = new OCRModelConfig();string modelPathroot = Path.Join(exePath,"ch_PP-OCRv4");config.det_infer = modelPathroot + @"\ch_PP-OCRv4_det_infer";config.cls_infer = modelPathroot + @"\ch_ppocr_mobile_v2.0_cls_infer";config.rec_infer = modelPathroot + @"\ch_PP-OCRv4_rec_infer";config.keys = modelPathroot + @"\ppocr_keys.txt";//初始化OCR引擎return new PaddleOCREngine(config, oCRParameter);}
}

Program.cs

using OCRTest02;
var exePath = AppDomain.CurrentDomain.BaseDirectory;
var engine = OCRHelper.initOcrEngine(exePath);
Console.WriteLine("初始化成功");
var filePath = "F:\\Desktop\\韦小宝.png";
var imagebyte = File.ReadAllBytes(filePath);
var ocrRes = engine.DetectText(imagebyte);foreach (var textBlock in ocrRes.TextBlocks)
{Console.Write(string.Join(";",textBlock.BoxPoints));Console.Write("\t"+textBlock.Text);Console.Write("\t"+textBlock.Score+"\n");
}

在这里插入图片描述
识别结果的精度还是比较高的但是目前数据比较乱，需要找到身份证号和姓名，下面代码可以识别横着、竖着的身份证信息并且提取出来，斜这的就不太行了，需要做图像校正，可以尝试一下https://github.com/shakex/card-rectification

using System.Text.RegularExpressions;
using OCRTest02;
using SixLabors.ImageSharp;
using SixLabors.ImageSharp.Drawing.Processing;
using SixLabors.ImageSharp.Processing;var exePath = AppDomain.CurrentDomain.BaseDirectory;
var engine = OCRHelper.initOcrEngine(exePath);
Console.WriteLine("初始化成功");
var filePath = "F:\\Desktop\\韦小宝.jpg";
var imagebyte = File.ReadAllBytes(filePath);
var ocrRes = engine.DetectText(imagebyte);string name = "";
string id = "";
var sfzImg = Image.Load(filePath);foreach (var textBlock in ocrRes.TextBlocks)
{//Console.WriteLine(textBlock.Text);#region 绘制包围矩形List<PointF> points = new List<PointF>();foreach (var point in textBlock.BoxPoints){points.Add(new PointF(point.X,point.Y));}sfzImg.Mutate(x=>x.DrawPolygon(new SolidPen(Color.HotPink,2.0f),points.ToArray()));#endregionif (string.IsNullOrWhiteSpace(textBlock.Text)){continue;}var tmpText = textBlock.Text.Trim();if (tmpText.Contains("姓名")){tmpText = tmpText.Replace("姓名", "");name = tmpText;}var idMatch = new Regex(@"^[1-9]\d{5}(18|19|([23]\d))\d{2}((0[1-9])|(10|11|12))(([0-2][1-9])|10|20|30|31)\d{3}[0-9Xx]$");if (idMatch.IsMatch(tmpText)){id = tmpText;}
}Console.WriteLine($"姓名:{name},身份证号:{id}");
var now = DateTime.Now;
var resultName = now.ToString("yyyy_MM_dd_HH_mm_ss");
sfzImg.Save($"{resultName}.jpg");
Console.WriteLine("保存成功");