[人工智能]OpenCVSharp 4.5 AKAZE 特征匹配

用?OpenCVSharp 4.5?跑一遍?OpenCV官方教程

原OpenCV官方教程链接：OpenCV: AKAZE local features matching

using System;
using System.Collections.Generic;
using OpenCvSharp;
using OpenCvSharp.XFeatures2D;

namespace ConsoleApp1
{
    class tutorial50:ITutorial
    {
        public void Run()
        {
            float inlier_threshold = 2.5f;
            float nn_match_ratio = 0.8f;

            Mat img1 = Cv2.ImRead(@"I:\csharp\images\graf1.png", ImreadModes.Grayscale);
            Mat img2 = Cv2.ImRead(@"I:\csharp\images\graf3.png",ImreadModes.Grayscale);
            
            //以下为 homography(文件H1to3p.xml)数据
            double[,] data = {{7.6285898e-01,-2.9922929e-01,2.2567123e+02 },
             { 3.3443473e-01, 1.0143901e+00, -7.6999973e+01 },
             { 3.4663091e-04, -1.4364524e-05, 1.0000000e+00 }};

            Mat homography = new Mat(3,3,MatType.CV_64F,data);
            KeyPoint[] kpts1, kpts2;
            Mat desc1 = new Mat(), desc2 = new Mat();

            AKAZE akaz = AKAZE.Create();
            akaz.DetectAndCompute(img1, null, out kpts1, desc1);
            akaz.DetectAndCompute(img2, null, out kpts2, desc2);

            BFMatcher matcher = new BFMatcher(NormTypes.Hamming);
            DMatch[][] nn_matches = matcher.KnnMatch(desc1, desc2, 2);

            List<KeyPoint> matched1 = new List<KeyPoint>(), matched2 = new List<KeyPoint>();
            for (int i = 0; i < nn_matches.Length; i++)
            {
                DMatch first = nn_matches[i][0];
                float dist1 = nn_matches[i][0].Distance;
                float dist2 = nn_matches[i][1].Distance;
                if (dist1 < nn_match_ratio * dist2)
                {
                    matched1.Add( kpts1[first.QueryIdx]);
                    matched2.Add(kpts2[first.TrainIdx]);
                }
            }
            List<DMatch> good_matches = new List<DMatch>();
            List<KeyPoint> inliers1=new List<KeyPoint>(),
                inliers2 = new List<KeyPoint>();
            for (int i = 0; i < matched1.Count; i++)
            {
                Mat col = Mat.Ones(new Size(1, 3), MatType.CV_64F);
                col.At<double>(0) = matched1[i].Pt.X;
                col.At<double>(1) = matched1[i].Pt.Y;
                col = homography * col;
                col /= col.At<double>(2);

                double dist = Math.Sqrt(Math.Pow(col.At<double>(0) - matched2[i].Pt.X, 2) +
                                        Math.Pow(col.At<double>(1) - matched2[i].Pt.Y, 2));
                if (dist < inlier_threshold)
                {
                    int new_i = (int)inliers1.Count;
                    inliers1.Add(matched1[i]);
                    inliers2.Add(matched2[i]);
                    good_matches.Add(new DMatch(new_i, new_i, 0));
                }
            }

            Mat res= new Mat();
            Cv2.DrawMatches(img1, inliers1, img2, inliers2, good_matches, res);
            Cv2.ImWrite("akaze_result.png", res);
            double inlier_ratio = inliers1.Count / (double)matched1.Count;
            Console.WriteLine( "A-KAZE Matching Results");
            Console.WriteLine("*******************************" );
            Console.WriteLine("# Keypoints 1:\t{0}" , kpts1.Length);
            Console.WriteLine("# Keypoints 2:\t{0}", kpts2.Length);
            Console.WriteLine("# Matches:\t{0}", matched1.Count);
            Console.WriteLine("# Inliers:\t{0}", inliers1.Count);
            Console.WriteLine("# Inliers Ratio:\t{0}" ,inlier_ratio );
            Cv2.ImShow("result", res);
            Cv2.WaitKey();
        }
    }
}

?

?