C# OpenCvSharp 基于直线检测的文本图像倾斜校正

效果

代码

复制代码
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Linq;
using System.Text;
using System.Windows.Forms;
using OpenCvSharp;
 
namespace OpenCvSharp_基于直线检测的文本图像倾斜校正
{
    public partial class Form1 : Form
    {
        public Form1()
        {
            InitializeComponent();
        }
 
        private void button1_Click(object sender, EventArgs e)
        {
            string path = "1.jpg";
 
            pictureBox1.Image = new Bitmap(path);
            Mat mat = new Mat(path);
 
            Mat gray = new Mat(path, ImreadModes.Grayscale);
 
            Mat binary = new Mat();
            Cv2.Threshold(gray, binary, 50, 255, ThresholdTypes.BinaryInv);
 
            Mat element = Cv2.GetStructuringElement(MorphShapes.Rect, new OpenCvSharp.Size(7, 1));
 
            Mat dilation = new Mat();
            Cv2.Dilate(binary, dilation, element);
 
            Mat cannyDst = new Mat();
            Cv2.Canny(dilation, cannyDst, 150, 200);
 
            Mat houghDst = new Mat();
            mat.CopyTo(houghDst);
 
            LineSegmentPolar[] lineing = Cv2.HoughLines(cannyDst, 1, Cv2.PI / 180, 110, 0, 0);
            Scalar color = new Scalar(0, 255, 255);
 
            double meanAngle = 0.0;
            int numCnt = 0;
 
            for (int i = 0; i < lineing.Length; i++)
            {
                double rho = lineing[i].Rho;//线长
                double theta = lineing[i].Theta;//角度
 
                OpenCvSharp.Point pt1 = new OpenCvSharp.Point();
                OpenCvSharp.Point pt2 = new OpenCvSharp.Point();
                double a = Math.Cos(theta);
                double b = Math.Sin(theta);
                double x0 = a * rho, y0 = b * rho;
 
                pt1.X = (int)Math.Round(x0 + 600 * (-b));
                pt1.Y = (int)Math.Round(y0 + 600 * a);
                pt2.X = (int)Math.Round(x0 - 600 * (-b));
                pt2.Y = (int)Math.Round(y0 - 600 * a);
 
                Cv2.Line(houghDst, pt1, pt2, color, 1, LineTypes.Link4);
 
                theta = theta * 180 / Cv2.PI - 90;
 
                meanAngle += theta;
                numCnt++;
            }
            //Cv2.ImShow("houghDst", houghDst);
 
            meanAngle /= numCnt;
            Point2f center = new Point2f(mat.Cols / 2.0f, mat.Rows / 2.0f);
 
            Mat warpDst = new Mat();
            Mat rot_mat = Cv2.GetRotationMatrix2D(center, meanAngle, 1.0);
            OpenCvSharp.Size dst_sz = new OpenCvSharp.Size(mat.Cols, mat.Rows);
 
            Cv2.WarpAffine(mat, warpDst, rot_mat, dst_sz);
 
            pictureBox2.Image = new Bitmap(warpDst.ToMemoryStream());
        }
    }
}

下载

Demo下载

相关推荐
mCell5 小时前
关于 Openclaw,最近的一点思考。
人工智能·安全·aigc
qq_171538856 小时前
纳采问名定佳期:中国传统订婚文化的千年传承与地域风华
人工智能
zzb15806 小时前
RAG from Scratch-优化-query
java·数据库·人工智能·后端·spring·mybatis
uzong6 小时前
315晚会曝光“AI大模型被投毒”,让AI听话,GEO是什么,带给我们什么思考
人工智能
V搜xhliang02466 小时前
机器人建模(URDF)与仿真配置
大数据·人工智能·深度学习·机器学习·自然语言处理·机器人
房产中介行业研习社6 小时前
2026年3月哪些房源管理系统功能全
大数据·运维·人工智能
Shining05966 小时前
CUDA 编程系列(三)《内存模型与规约优化》
人工智能·学习·其他·学习方法·infinitensor
lisw057 小时前
基于图像的恶意软件分类方法!
人工智能·机器学习
L-影7 小时前
AI中的Transformer:从RNN的困境到横扫一切的革命(下篇)
人工智能·rnn·ai·transformer
卷福同学7 小时前
QClaw内测体验,能用微信指挥AI干活了
人工智能·算法·ai编程