在前两篇文章中，我介绍了《训练自己的haar-like特征分类器并识别物体》的前三个步骤：
1.准备训练样本图片，包括正例及反例样本
2.生成样本描述文件
3.训练样本
4.目标识别
==============
本文将着重说明最后一个阶段——目标识别，也即利用前面训练出来的分类器文件(.xml文件)对图片中的物体进行识别，并在图中框出在该物体。由于逻辑比较简单，这里直接上代码：
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
|
int
_tmain(
int
argc, _TCHAR* argv[])
{
char
*cascade_name = CASCADE_HEAD_MY;
//上文最终生成的xml文件命名为"CASCADE_HEAD_MY.xml"
cascade = (CvHaarClassifierCascade*)cvLoad( cascade_name, 0, 0, 0 );
//加载xml文件
if
( !cascade )
{
fprintf
( stderr,
"ERROR: Could not load classifier cascade\n"
);
system
(
"pause"
);
return
-1;
}
storage = cvCreateMemStorage(0);
cvNamedWindow(
"face"
, 1 );
const
char
* filename =
"(12).bmp"
;
IplImage* image = cvLoadImage( filename, 1 );
if
( image )
{
detect_and_draw( image );
//函数见下方
cvWaitKey(0);
cvReleaseImage( &image );
}
cvDestroyWindow(
"result"
);
return
0;
}
|
1 void detect_and_draw(IplImage* img ) 2 { 3 double scale=1.2; 4 static CvScalar colors[] = { 5 {{0,0,255}},{{0,128,255}},{{0,255,255}},{{0,255,0}}, 6 {{255,128,0}},{{255,255,0}},{{255,0,0}},{{255,0,255}} 7 };//Just some pretty colors to draw with 8 9 //Image Preparation 10 // 11 IplImage* gray = cvCreateImage(cvSize(img->width,img->height),8,1); 12 IplImage* small_img=cvCreateImage(cvSize(cvRound(img->width/scale),cvRound(img->height/scale)),8,1); 13 cvCvtColor(img,gray, CV_BGR2GRAY); 14 cvResize(gray, small_img, CV_INTER_LINEAR); 15 16 cvEqualizeHist(small_img,small_img); //直方图均衡 17 18 //Detect objects if any 19 // 20 cvClearMemStorage(storage); 21 double t = (double)cvGetTickCount(); 22 CvSeq* objects = cvHaarDetectObjects(small_img, 23 cascade, 24 storage, 25 1.1, 26 2, 27 0/*CV_HAAR_DO_CANNY_PRUNING*/, 28 cvSize(30,30)); 29 30 t = (double)cvGetTickCount() - t; 31 printf( "detection time = %gms\n", t/((double)cvGetTickFrequency()*1000.) ); 32 33 //Loop through found objects and draw boxes around them 34 for(int i=0;i<(objects? objects->total:0);++i) 35 { 36 CvRect* r=(CvRect*)cvGetSeqElem(objects,i); 37 cvRectangle(img, cvPoint(r->x*scale,r->y*scale), cvPoint((r->x+r->width)*scale,(r->y+r->height)*scale), colors[i%8]); 38 } 39 for( int i = 0; i < (objects? objects->total : 0); i++ ) 40 { 41 CvRect* r = (CvRect*)cvGetSeqElem( objects, i ); 42 CvPoint center; 43 int radius; 44 center.x = cvRound((r->x + r->width*0.5)*scale); 45 center.y = cvRound((r->y + r->height*0.5)*scale); 46 radius = cvRound((r->width + r->height)*0.25*scale); 47 cvCircle( img, center, radius, colors[i%8], 3, 8, 0 ); 48 } 49 50 cvShowImage( "result", img ); 51 cvReleaseImage(&gray); 52 cvReleaseImage(&small_img); 53 }
===================================
其实上面的代码可以运用于大部分模式识别问题，无论是自己生成的xml文件还是opencv自带的xml文件。在opencv的工程目录opencv\data文件夹下有大量的xml文件，这些都是opencv开源项目中的程序员们自己训练出来的。然而，效果一般不会合你预期，所以才有了本系列文章。天下没有免费的午餐，想要获得更高的查准率与查全率，不付出点努力是不行的！
