opencv-自定義harris角點檢測
阿新 • • 發佈:2019-03-23
程序 number 是什麽 val input lsa key oid tps
opencv-自定義harris角點檢測
關於harris角點檢測的原理以及matlab版本,請移步https://www.cnblogs.com/klitech/p/5779600.html
小白初學,這裏采用opencv實現之,把自己遇到的疑問一一表述出來,以備後用。
疑問1. CV_32FC1,CV_32FC(6)是什麽意思?
CV_<bit_depth>(S|U|F)C<number_of_channels>
-
- bit_depth,可取的值為8,16,32,64.它表示一個像素所占用的bite數
- S|U|F, S--- signed int, U--- unsigned int, F---float
- C<number_of_channels> 通道數,灰度圖像取值1,rgb圖像取值3
疑問2. Mat::zeros的兩種初始化
- Mat::zeros(int rows, int cols, int type)
- Mat::zeros(Size size,int type)
第一種形式,返回特定尺寸與類型的零矩陣,比如 Mat A = Mat::zeros(3,3,CV_32FC1);
第二種形式,程序中采用的方式,Mat::zeros(src.size(), CV_32FC(6));
我的理解是這兩種形式實質一樣
疑問3. cornerEigenValsAndVecs()使用方法
函數原型, cornerEigenValsAndVecs( InputArray src, OutputArray dst,
int blockSize, int ksize,
int borderType = BORDER_DEFAULT );
- src 圖像類型應該為單通道,或者float
- dst 圖像類型應該為CV_32FC(6),包含2個特征值,以及對應的2個2維向量,總計6個結果。
- blocksize 鄰域大小
- ksize 函數采用sobel算子
- borderType 取默認BORDER_DEFAULT
函數調用參看後面的程序。
1 #include <opencv2/opencv.hpp> 2 #include <iostream> 3 #include <math.h> 4 using namespace cv; 5 using namespace std; 6 Mat src, gray_src; 7 Mat harris_dst, harrisRspImg; 8 double harris_min_rsp; 9 double harris_max_rsp; 10 int qualityLevel = 30; 11 const char* harris_win = "Custom Harris Corners Dector"; 12 int max_count = 100; 13 void CustomHarris_Demo(int, void *); 14 15 int main() 16 { 17 18 src = imread("D:/1.png"); 19 if (src.empty()) 20 { 21 cout << "could not load image..." << endl; 22 return -1; 23 } 24 namedWindow("input_image", CV_WINDOW_AUTOSIZE); 25 imshow("input_image", src); 26 cvtColor(src, gray_src, COLOR_BGR2GRAY); 27 // 計算特征值 28 int blockSize = 3; 29 int ksize = 3; 30 double k = 0.04; 31 32 harris_dst = Mat::zeros(src.size(), CV_32FC(6)); //6通道 33 harrisRspImg = Mat::zeros(src.size(), CV_32FC1); 34 cornerEigenValsAndVecs(gray_src, harris_dst, blockSize, ksize, 4); 35 //計算響應 36 for (int row = 0; row < harris_dst.rows; row++) 37 { 38 for (int col = 0; col < harris_dst.cols; col++) 39 { 40 double lambda1 = harris_dst.at<Vec6f>(row, col)[0]; 41 double lambda2 = harris_dst.at<Vec6f>(row, col)[1]; 42 harrisRspImg.at<float>(row, col) = lambda1 * lambda2 - k*pow((lambda1 + lambda2), 2); 43 } 44 } 45 minMaxLoc(harrisRspImg, &harris_min_rsp, &harris_max_rsp, 0, 0, Mat());//求最大最小響應 46 namedWindow(harris_win, CV_WINDOW_AUTOSIZE); 47 createTrackbar("Quality Value", harris_win, &qualityLevel, max_count, CustomHarris_Demo); 48 CustomHarris_Demo(0, 0); 49 waitKey(0); 50 return 0; 51 } 52 void CustomHarris_Demo(int, void*) { 53 if (qualityLevel < 10) { 54 qualityLevel = 10; 55 } 56 Mat resultImg = src.clone(); 57 float t = harris_min_rsp + (((double)qualityLevel) / max_count)*(harris_max_rsp - harris_min_rsp); 58 for (int row = 0; row < src.rows; row++) { 59 for (int col = 0; col < src.cols; col++) { 60 float v = harrisRspImg.at<float>(row, col); 61 if (v > t) { 62 circle(resultImg, Point(col, row), 2, Scalar(0, 0, 255), 2, 8, 0); 63 } 64 } 65 } 66 67 imshow(harris_win, resultImg); 68 }
opencv-自定義harris角點檢測