[learning opencv]第十章cvGoodFeaturesToTrack的demo

從標題看，相關內容可參閱《learning opencv》

本文摘自

http://blog.csdn.net/moc062066/article/details/6634120

void cvGoodFeaturesToTrack(
const CvArr* image //(8,1) or (32,1) (8-bit ,single-channel) (floating-point 32-bit,single-channel)
CvArr* eigImage,//(32,1)
CvArr* tempImage //(32,1)
CvPoint2D32f* corners
int* cornerCount
double qualityLevel
double minDistance
const CvArr* mask=NULL
int blockSize=3
int useHarris=0
double k=0.04 );

corners //是數組,檢測到的角點的位置座標就存在這裏

int* cornerCount //表示最多可以檢測到的角點（如果有那麼多的話），實際上檢測到的點沒有那麼多，所以調用函數以後，<pre name="code" class="cpp">cornerCount的值就變成實際檢測到的角點的數目。因此，在對cornerCount賦初始值的時候，可以給大一點，本文賦初始值爲1000。</pre>

double qualityLevel //理解還不是很好，只知道一般取值是0.10 或者0.01 ，why？誰可以告訴我答案

minDistance Limit, specifying the minimum possible distance between the returned corners;
Euclidian distance is used //角點與角點之間的距離不小於minDistance個像素。

mask Region of interest. The function selects points either in the specified region or in the whole
image if the mask is NULL //如果mask不爲空，則在mask指定的區域內尋找角點；mask爲NULL則在整張圖像中尋找角點。

blockSize Size of the averaging block, passed to the underlying cvCornerMinEigenVal or cvCornerHarris used by the function //默認就行了

useHarris If nonzero, Harris operator ( cvCornerHarris) is used instead of default cvCorner-
MinEigenVal //非零，就用cvCornerHarris而不是使用默認的cvCornerMinEigenVal。這是函數內部的處理過程。

k Free parameter of Harris detector; used only if (useHarris! = 0)

demo代碼如下，不解析，很簡單

//http://blog.csdn.net/moc062066
//[email protected]
//2011年7月26日10:41:54
//

#include <cv.h>
#include <highgui.h>
#include <stdio.h>

#pragma comment(lib, "opencv_core220d.lib")
#pragma comment(lib, "opencv_highgui220d.lib")
#pragma comment(lib, "opencv_imgproc220d.lib")

int main(int argc, char *argv[])
{

    // Load a color image, and convert it into grayscale
    const char* filename = "D:\\mochen_WIN32\\opencv\\CH10\\good_fearture_to_track_demo\\one_way_train_0001.jpg" ;
    const char* windowname = "http://blog.csdn.net/moc062066" ;

    IplImage* img = cvLoadImage(filename,CV_LOAD_IMAGE_COLOR);
    assert( NULL != img ) ;
    IplImage* img_gray = cvCreateImage(cvGetSize(img), 8, 1);
    cvCvtColor(img, img_gray, CV_BGR2GRAY);

    // Create temporary images required by cvGoodFeaturesToTrack
    IplImage* img_temp = cvCreateImage(cvGetSize(img), 32, 1);
    IplImage* img_eigen = cvCreateImage(cvGetSize(img), 32, 1);

    // Create the array to store the points detected( <= 1000 )
    int count = 1000;
    CvPoint2D32f* corners = new CvPoint2D32f[count];

    // Find corners
    cvGoodFeaturesToTrack(img_gray, img_eigen, img_temp, corners, &count, 0.1, 10);

    // Mark these corners on the original image
    for(int i=0;i<count;i++)
    {
        cvLine(img, cvPoint(corners[i].x, corners[i].y), cvPoint(corners[i].x, corners[i].y), CV_RGB(255,0,0), 5);
    }

    // Display it
    cvNamedWindow(windowname);
    cvShowImage(windowname, img);

    // Print the number of corners
    printf("Detected Points : %d\n", count);

    cvWaitKey(0);

    return  0 ;
}

結果：

實驗所用圖像：