Record
1、特征点检测与匹配常用的算法:FAST(FastFeatureDetector)、STAR(StarFeatureDetector)、SIFT、SURF、ORB、MSER、GFTT(GoodFeaturesToTrackDetector)、HARRIS、Dense、SimpleBlob等;
2、在EmguCV中,SIFT与SURF位于命名空间Emgu.CV.XFeatures2D下;
3、SIFT—尺度不变特征变换检测算法,SIFT特征对旋转,尺度缩放,亮度变化等保持不变性,是非常稳定的局部特征,应用广泛;
4、SIFT原理:设置尺度空间滤波器,关键点定位,为关键点指定方向参数(保持旋转不变性),
5、SIFT算法接口:
MKeyPoint结构是用于表示特征点:
DrawKeyPoints()函数用于绘制所有关键点:
Code
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using Emgu.CV;
using Emgu.CV.CvEnum;
using Emgu.Util;
using Emgu.CV.Structure;
using Emgu.CV.XFeatures2D;
using System.Drawing;
using Emgu.CV.Features2D;
using Emgu.CV.Util;
namespace siftDetect
{
class Program
{
static void Main(string[] args)
{
Mat src1 = CvInvoke.Imread("01.jpg");
//Mat src1 = CvInvoke.Imread("00.jpg");
SIFT sift = new SIFT(1000,3,0.04,10,1.6);
MKeyPoint[] mKeyPoints = sift.Detect(src1);
Mat sift_feature = new Mat();
VectorOfKeyPoint vkPoints = new VectorOfKeyPoint(mKeyPoints);
Features2DToolbox.DrawKeypoints(src1, vkPoints, sift_feature,
new Bgr(0, 255, 0), Features2DToolbox.KeypointDrawType.Default);
Random random = new Random();
for (int i = 0; i < mKeyPoints.Length; i++)
{
Point keyP = new Point();
keyP.X = (int)mKeyPoints[i].Point.X;
keyP.Y = (int)mKeyPoints[i].Point.Y;
CvInvoke.Circle(src1, keyP, 3, new MCvScalar(random.Next(0, 255), random.Next(0, 255), random.Next(0, 255)), -1);
}
CvInvoke.Imshow("drawkeypoint", sift_feature);
CvInvoke.Imshow("out", src1);
CvInvoke.WaitKey(0);
}
}
}
效果
该算法计算的特征点具有旋转不变性与缩放不变性