Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 10 months ago.
Improve this question
**Requirements: **
(1) Build OpenCV with Cuda and compile in C++
(2) Version = OpenCV Latest Version
(3) Build and Compile OpenCV Link: https://techawarey.com/programming/install-opencv-c-c-in-ubuntu-18-04-lts-step-by-step-guide/#Summary
(4) Library samples_utility is here: https://github.com/opencv/opencv_contrib/blob/master/modules/tracking/samples/samples_utility.hpp
(4) Compile Program Command: g++ test.cpp -o testoutput -std=c++11 'pkg-config --cflags --libs opencv'
(5) Run Program Command: ./testoutput
Code is working fine but not accurate
Step: 1
Read Frame from Camera
Select ROI(Region of Interest)
After that start KCF tracker with Sobal Features Extractor
Tracking the selected object.
Step: 2
Failure detect
After that call template matching function called MatchingMethod()
Run template matching
Get x, y value from template matching
After that reinitialize KCF tracker with Sobal Features Extractor.
This code is fine for still object when the object is moving the tracker false detection. I want to improve accuracy and reduce false detection.
#include <opencv2/core/utility.hpp>
#include <opencv2/tracking.hpp>
#include <opencv2/videoio.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/core/ocl.hpp>
#include <iostream>
#include <cstring>
#include <unistd.h>
#include "sample_utility.hpp"
#include <thread>
#include <opencv2/cudaimgproc.hpp>
#include <opencv2/imgproc.hpp>
#include <opencv2/calib3d.hpp>
#include <opencv2/cudaarithm.hpp>
#include <iomanip>
#include <stdlib.h>
#include <unistd.h>
//////////////////////////////
using namespace cv;
using namespace std;
////////////////////////////
// Convert to string
#define SSTR( x ) static_cast< std::ostringstream & >( \
( std::ostringstream() << std::dec << x ) ).str()
/// Global Variables
struct Array {
int arr[2];
};
Mat img;
Mat templ;
Mat result_h;
bool flag = true;
int match_method = 5;
int i=0;
int max_Trackbar = 5;
float fps;
int seconds = 0;
// Function Headers
void delay();
// prototype of the functino for feature extractor
void sobelExtractor(const Mat img, const Rect roi, Mat& feat);
struct Array MatchingMethod( int, void* );
int main(int argc, char **argv)
{
TrackerKCF::Params param;
param.compress_feature = true;
param.compressed_size = 2;
param.desc_npca = 0;
param.desc_pca = TrackerKCF::GRAY | TrackerKCF::CN;
param.detect_thresh = 0.8;
// create a tracker object
Ptr<TrackerKCF> tracker = TrackerKCF::create(param);
tracker->setFeatureExtractor(sobelExtractor);
VideoCapture cap(0);
// Exit if video is not opened
if(!cap.isOpened())
{
//cout << "Could not read video file" << endl;
return 1;
}
// Read first frame
Mat frame;
bool ok = cap.read(frame);
// Define initial bounding box
//Rect bbox(x, y, w, h);
// Uncomment the line below to select a different bounding box
Rect bbox = selectROI(frame, false);
// Display bounding box.
rectangle(frame, bbox, Scalar( 255, 0, 0 ), 2, 1 );
///////////////////////////
int H, W, cW, cH;
// print(f"hight {H} , Width {W}")
H = display_height;
W = display_width;
// Center point of the screen
cW = int(W / 2);
cH = int(H / 2);
Point p1(cW, cH);
// get bounding box
Mat imCrop = frame(bbox);
imwrite("1.png", imCrop);
//quit if ROI was not selected
if(bbox.width==0 || bbox.height==0)
return 0;
//////////////////////////
//imshow("Tracking", frame);
tracker->init(frame, bbox);
while(true)
{
Mat frame;
cap >> frame;
circle(frame, p1, 3, Scalar(0,255,0), -1);
// Start timer
if(bbox.width!=0 || bbox.height!=0){
double timer = (double)getTickCount();
// Update the tracking result
/////////////////////////////////////
bool ok = tracker->update(frame, bbox);
//////////////////////////////////////
//ok, bbox = tracker->update(frame);
// Calculate Frames per second (FPS)
fps = getTickFrequency() / ((double)getTickCount() - timer);
if (ok)
{
// Tracking success : Draw the tracked object
rectangle(frame, bbox, Scalar( 255, 0, 0 ), 2, 1 );
///////////////////////////////////////////////////
int xxx, yyy, height, width;
xxx = bbox.x;
yyy = bbox.y;
height = bbox.height;
width = bbox.width;
int diffX, diffY;
float cxROI, cyROI;
cxROI = int((xxx + (xxx + width)) / 2);
cyROI = int((yyy + (yyy + height)) / 2);
diffX = cxROI - cW;
diffY = cH - cyROI;
//cout<<diffX<<"\n";
//cout<<diffY<<"\n";
Point p(cxROI, cyROI);
circle(frame, p, 3, Scalar(128,0,0), -1);
putText(frame, "FPS : " + SSTR(int(fps)), Point(100,20), FONT_HERSHEY_SIMPLEX, 0.75, Scalar(50,170,50), 2);
putText(frame, "Difference From X-Axis: "+SSTR(int(diffX)), Point(100, 50), FONT_HERSHEY_SIMPLEX, 0.6, Scalar(100, 200, 200), 2);
putText(frame, "Difference From Y-Axis: "+SSTR(int(diffY)), Point(100, 80), FONT_HERSHEY_SIMPLEX, 0.6, Scalar(100, 200, 200), 2);
}
else
{
// Tracking failure detected.
putText(frame, "Tracking failure detected", Point(100,110), FONT_HERSHEY_SIMPLEX, 0.75, Scalar(0,0,255),2);
templ = imread( "1.png", 1 );
img=frame.clone();
struct Array a = MatchingMethod( 0, 0 );
cout<<"X: "<<a.arr[0]<<"\n";
cout<<"Y: "<<a.arr[1]<<"\n";
cout<<"Width: "<<w<<"\n";
cout<<"Height: "<<h<<"\n";
int xx, yy, ww, hh;
xx = a.arr[0];
yy = a.arr[1];
ww = w;
hh = h;
Rect bbox(xx, yy, ww, hh);
tracker.release();
tracker = TrackerKCF::create(param);
tracker->setFeatureExtractor(sobelExtractor);
tracker->init(frame, bbox);
//roi.x = MatchingMethod.
//waitKey(30);
rectangle(frame, bbox, Scalar( 255, 0, 0 ), 2, 1 );
////////////////////////////////////////////////////////////////////////
int diffX, diffY;
float cxROI, cyROI;
cxROI = int((xx + (xx + ww)) / 2);
cyROI = int((yy + (yy + hh)) / 2);
diffX = cxROI - cW;
diffY = cH - cyROI;
//cout<<diffX<<"\n";
//cout<<diffY<<"\n";
Point p(cxROI, cyROI);
circle(frame, p, 3, Scalar(128,0,0), -1);
///////////////////////////////////////////////////////////////////////////
}
}
else{
}
// Display frame.
imshow("Tracking", frame);
// Exit if ESC pressed.
int k = waitKey(1);
if(k == 27)
{
break;
}
}
return 0;
}
///////////////////////
void sobelExtractor(const Mat img, const Rect roi, Mat& feat){
Mat sobel[2];
Mat patch;
Rect region=roi;
// extract patch inside the image
if(roi.x<0){region.x=0;region.width+=roi.x;}
if(roi.y<0){region.y=0;region.height+=roi.y;}
if(roi.x+roi.width>img.cols)region.width=img.cols-roi.x;
if(roi.y+roi.height>img.rows)region.height=img.rows-roi.y;
if(region.width>img.cols)region.width=img.cols;
if(region.height>img.rows)region.height=img.rows;
patch=img(region).clone();
cvtColor(patch,patch, COLOR_BGR2GRAY);
// add some padding to compensate when the patch is outside image border
int addTop,addBottom, addLeft, addRight;
addTop=region.y-roi.y;
addBottom=(roi.height+roi.y>img.rows?roi.height+roi.y-img.rows:0);
addLeft=region.x-roi.x;
addRight=(roi.width+roi.x>img.cols?roi.width+roi.x-img.cols:0);
copyMakeBorder(patch,patch,addTop,addBottom,addLeft,addRight,BORDER_REPLICATE);
Sobel(patch, sobel[0], CV_32F,1,0,1);
Sobel(patch, sobel[1], CV_32F,0,1,1);
merge(sobel,2,feat);
feat=feat/255.0-0.5; // normalize to range -0.5 .. 0.5
}
////////////////////////////////////////////////////
struct Array MatchingMethod( int, void* )
{
/// Source image to display
Mat frame;
struct Array a;
/////////
for(int i=1; i<=4; i++){
img.copyTo( frame );
// break;
//}
//////////////////////////
cv::cuda::setDevice(0); // initialize CUDA
// convert from mat to gpumat
cv::cuda::GpuMat image_d(img);
cv::cuda::GpuMat templ_d(templ);
cv::cuda::GpuMat result;
// GPU -> NG
cv::Ptr<cv::cuda::TemplateMatching> alg =
cv::cuda::createTemplateMatching(image_d.type(), cv::TM_CCOEFF_NORMED);
alg->match(image_d, templ_d, result); // no return.
//cv::cuda::normalize(result, result, 0, 1, cv::NORM_MINMAX, -1);
double max_value, min_value;
cv::Point location;
cv::cuda::minMaxLoc(result, &min_value, &max_value, 0, &location);
/////////////////////////
double THRESHOLD = 3e-09; //0.3;
if( min_value <= THRESHOLD) {
//struct Array a;
a.arr[0] = location.x;
a.arr[1] = location.y;
cout<<"Hi"<<endl;
}
}
if(flag==true){
return a;
flag = false;
}
//}
}
Okey here is my answer to your question.
First of all, you are making a mistake by applying template matching when the tracker misses. Because template matching matches the feature if and only if it is totally same with the reference feature. So in your case, there will be shadows, light issues etc. in the environment, and you will never be able to get success results.
Secondly, if you delete the template matching scope, tracker will continue to search the target in the image effectively. Which changements I did in your code is listed below. With these changes, I got better results:
Delete the template matching scope
Decrease the detection threshold(param.detect_thresh) to 0.5
Create more tracker objects to catch the target: This change is the most important part. What I am suggesting is that create more and more tracker objects(in my case I did 4 tracker objects, but you can increase the number). Each tracker should get as input rectangle similar to ROI user chose but not the same coordinates. For example, if user chooses cv::Rect(200,200,400,400) then other tracker should get target as cv::Rect(180,190,400,400) , cv::Rect(220,180,400,400) ... and so on. Why you should do it because, tracker algorithm is feature based, so it will always try to get a similar features to the reference. By doing this, you will increase the feature references.
And here is my code to guide you:
#include <opencv2/core/utility.hpp>
#include <opencv2/tracking.hpp>
#include <opencv2/videoio.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/core/ocl.hpp>
#include <iostream>
#include <cstring>
#include <unistd.h>
#include <thread>
#include <opencv2/cudaimgproc.hpp>
#include <opencv2/imgproc.hpp>
#include <opencv2/calib3d.hpp>
#include <opencv2/cudaarithm.hpp>
#include <iomanip>
#include <stdlib.h>
#include <unistd.h>
//////////////////////////////
using namespace cv;
using namespace std;
////////////////////////////
// Convert to string
#define SSTR( x ) static_cast< std::ostringstream & >( \
( std::ostringstream() << std::dec << x ) ).str()
/// Global Variables
struct Array {
int arr[2];
};
Mat img;
Mat templ;
Mat result_h;
bool flag = true;
int match_method = 5;
int i=0;
int max_Trackbar = 5;
float fps;
int seconds = 0;
// Function Headers
void delay();
// prototype of the functino for feature extractor
void sobelExtractor(const Mat img, const Rect roi, Mat& feat);
struct Array MatchingMethod( int, void* );
int main(int argc, char **argv)
{
TrackerKCF::Params param;
param.compress_feature = true;
param.compressed_size = 2;
param.desc_npca = 0;
param.desc_pca = TrackerKCF::GRAY | TrackerKCF::CN;
param.detect_thresh = 0.5;
// create a tracker object
Ptr<TrackerKCF> tracker = TrackerKCF::create(param);
tracker->setFeatureExtractor(sobelExtractor);
Ptr<TrackerKCF> tracker2 = TrackerKCF::create(param);
tracker2->setFeatureExtractor(sobelExtractor);
Ptr<TrackerKCF> tracker3 = TrackerKCF::create(param);
tracker3->setFeatureExtractor(sobelExtractor);
Ptr<TrackerKCF> tracker4 = TrackerKCF::create(param);
tracker4->setFeatureExtractor(sobelExtractor);
VideoCapture cap(0);
// Exit if video is not opened
if(!cap.isOpened())
{
//cout << "Could not read video file" << endl;
return 1;
}
cv::imshow("Tracking",0);
// Read first frame
Mat frame;
bool ok = cap.read(frame);
// Define initial bounding box
//Rect bbox(x, y, w, h);
// Uncomment the line below to select a different bounding box
Rect2d bbox = selectROI(frame, false);
// Display bounding box.
rectangle(frame, bbox, Scalar( 255, 0, 0 ), 2, 1 );
///////////////////////////
int H, W, cW, cH;
// print(f"hight {H} , Width {W}")
H = frame.rows;
W = frame.cols;
// Center point of the screen
cW = int(W / 2);
cH = int(H / 2);
Point p1(cW, cH);
//quit if ROI was not selected
if(bbox.width==0 || bbox.height==0)
return 0;
//////////////////////////
//imshow("Tracking", frame);
tracker->init(frame, bbox);
tracker2->init(frame, cv::Rect2d(bbox.x-10,bbox.y-10, bbox.width,bbox.height));
tracker3->init(frame, cv::Rect2d(bbox.x+10,bbox.y+10, bbox.width,bbox.height));
tracker4->init(frame, cv::Rect2d(bbox.x+20,bbox.y+20, bbox.width,bbox.height));
while(true)
{
Mat frame;
cap >> frame;
circle(frame, p1, 3, Scalar(0,255,0), -1);
// Start timer
if(bbox.width!=0 || bbox.height!=0){
double timer = (double)getTickCount();
// Update the tracking result
/////////////////////////////////////
bool ok = tracker->update(frame, bbox);
bool ok2 = tracker->update(frame, bbox);
bool ok3 = tracker->update(frame, bbox);
bool ok4 = tracker->update(frame, bbox);
//////////////////////////////////////
//ok, bbox = tracker->update(frame);
// Calculate Frames per second (FPS)
fps = getTickFrequency() / ((double)getTickCount() - timer);
if (ok || ok2 || ok3 || ok4)
{
// Tracking success : Draw the tracked object
rectangle(frame, bbox, Scalar( 255, 0, 0 ), 2, 1 );
///////////////////////////////////////////////////
int xxx, yyy, height, width;
xxx = bbox.x;
yyy = bbox.y;
height = bbox.height;
width = bbox.width;
int diffX, diffY;
float cxROI, cyROI;
cxROI = int((xxx + (xxx + width)) / 2);
cyROI = int((yyy + (yyy + height)) / 2);
diffX = cxROI - cW;
diffY = cH - cyROI;
//cout<<diffX<<"\n";
//cout<<diffY<<"\n";
Point p(cxROI, cyROI);
circle(frame, p, 3, Scalar(128,0,0), -1);
putText(frame, "FPS : " + SSTR(int(fps)), Point(100,20), FONT_HERSHEY_SIMPLEX, 0.75, Scalar(50,170,50), 2);
putText(frame, "Difference From X-Axis: "+SSTR(int(diffX)), Point(100, 50), FONT_HERSHEY_SIMPLEX, 0.6, Scalar(100, 200, 200), 2);
putText(frame, "Difference From Y-Axis: "+SSTR(int(diffY)), Point(100, 80), FONT_HERSHEY_SIMPLEX, 0.6, Scalar(100, 200, 200), 2);
}
}
else{
}
// Display frame.
imshow("Tracking", frame);
// Exit if ESC pressed.
int k = waitKey(1);
if(k == 27)
{
break;
}
}
return 0;
}
///////////////////////
void sobelExtractor(const Mat img, const Rect roi, Mat& feat){
Mat sobel[2];
Mat patch;
Rect region=roi;
// extract patch inside the image
if(roi.x<0){region.x=0;region.width+=roi.x;}
if(roi.y<0){region.y=0;region.height+=roi.y;}
if(roi.x+roi.width>img.cols)region.width=img.cols-roi.x;
if(roi.y+roi.height>img.rows)region.height=img.rows-roi.y;
if(region.width>img.cols)region.width=img.cols;
if(region.height>img.rows)region.height=img.rows;
patch=img(region).clone();
cvtColor(patch,patch, COLOR_BGR2GRAY);
// add some padding to compensate when the patch is outside image border
int addTop,addBottom, addLeft, addRight;
addTop=region.y-roi.y;
addBottom=(roi.height+roi.y>img.rows?roi.height+roi.y-img.rows:0);
addLeft=region.x-roi.x;
addRight=(roi.width+roi.x>img.cols?roi.width+roi.x-img.cols:0);
copyMakeBorder(patch,patch,addTop,addBottom,addLeft,addRight,BORDER_REPLICATE);
Sobel(patch, sobel[0], CV_32F,1,0,1);
Sobel(patch, sobel[1], CV_32F,0,1,1);
merge(sobel,2,feat);
feat=feat/255.0-0.5; // normalize to range -0.5 .. 0.5
}
////////////////////////////////////////////////////
struct Array MatchingMethod( int, void* )
{
/// Source image to display
Mat frame;
struct Array a;
/////////
for(int i=1; i<=4; i++){
img.copyTo( frame );
// break;
//}
//////////////////////////
cv::cuda::setDevice(0); // initialize CUDA
// convert from mat to gpumat
cv::cuda::GpuMat image_d(img);
cv::cuda::GpuMat templ_d(templ);
cv::cuda::GpuMat result;
// GPU -> NG
cv::Ptr<cv::cuda::TemplateMatching> alg =
cv::cuda::createTemplateMatching(image_d.type(), cv::TM_CCOEFF_NORMED);
alg->match(image_d, templ_d, result); // no return.
//cv::cuda::normalize(result, result, 0, 1, cv::NORM_MINMAX, -1);
double max_value, min_value;
cv::Point location;
cv::cuda::minMaxLoc(result, &min_value, &max_value, 0, &location);
/////////////////////////
double THRESHOLD = 3e-09; //0.3;
if( min_value <= THRESHOLD) {
//struct Array a;
a.arr[0] = location.x;
a.arr[1] = location.y;
cout<<"Hi"<<endl;
}
}
if(flag==true){
return a;
flag = false;
}
//}
}
Could anyone please help me write the output video file? I have read many similar questions on how to write the program and followed the exact steps to write the video file in .avi format, but I am not able to find out where I am wrong. The face_output.avi file is created but it only contains one frame. My program is not adding all the frames to the video file. Below is the complete code:
#include "opencv2/objdetect/objdetect.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include <iostream>
#include <stdio.h>
using namespace std;
using namespace cv;
/** Function Headers */
void detectAndDisplay( Mat frame);
/** Global variables */
String face_cascade_name = "haarcascade_frontalface_alt.xml";
String eyes_cascade_name = "haarcascade_eye_tree_eyeglasses.xml";
CascadeClassifier face_cascade;
CascadeClassifier eyes_cascade;
string window_name = "Capture - Face detection";
RNG rng(12345);
double fps;
/** #function main */
int main( int argc, const char** argv )
{
VideoCapture cap("/home/pradeep/Downloads/President Obama Lectures Romney.mp4"); // open the video file for reading
if ( !cap.isOpened() ) // if not success, exit program
{
cout << "Cannot open the video file" << endl;
return -1;
}
//-- 1. Load the cascades
if( !face_cascade.load( face_cascade_name ) ){ printf("--(!)Error loading\n"); return -1; };
if( !eyes_cascade.load( eyes_cascade_name ) ){ printf("--(!)Error loading\n"); return -1; };
fps = cap.get(CV_CAP_PROP_FPS); //get the frames per seconds of the video
cout << "Frame per seconds : " << fps << endl;
double dWidth = cap.get(CV_CAP_PROP_FRAME_WIDTH);
double dHeight = cap.get(CV_CAP_PROP_FRAME_HEIGHT);
Size S(dWidth,dHeight);
while(1)
{
Mat frame;
int skip_frame = 4;
while(skip_frame)
{
bool bSuccess = cap.read(frame); // read a new frame from video
skip_frame--;
if (!bSuccess) //if not success, break loop
{
cout << "Cannot read the frame from video file" << endl;
break;
}
}
//-- 3. Apply the classifier to the frame
if( frame.empty() )
{ printf(" --(!) No captured frame -- Break!"); break; }
std::vector<Rect> faces;
Mat frame_gray;
cvtColor( frame, frame_gray, CV_BGR2GRAY );
equalizeHist( frame_gray, frame_gray );
//-- Detect faces
face_cascade.detectMultiScale( frame_gray, faces, 1.3, 5, 0|CV_HAAR_SCALE_IMAGE, Size(30, 30) );
for( size_t i = 0; i < faces.size(); i++ )
{
Point center( faces[i].x + faces[i].width*0.5, faces[i].y + faces[i].height*0.5 );
ellipse( frame, center, Size( faces[i].width*0.5, faces[i].height*0.5), 0, 0, 360, Scalar( 255, 0, 255 ), 4, 8, 0 );
Mat faceROI = frame_gray( faces[i] );
std::vector<Rect> eyes;
//-- In each face, detect eyes
eyes_cascade.detectMultiScale( faceROI, eyes, 1.1, 0, 0 |CV_HAAR_SCALE_IMAGE, Size(30, 30) );
for( size_t j = 0; j < eyes.size(); j++ )
{
Point center( faces[i].x + eyes[j].x + eyes[j].width*0.5, faces[i].y + eyes[j].y + eyes[j].height*0.5 );
int radius = cvRound( (eyes[j].width + eyes[j].height)*0.25 );
circle( frame, center, radius, Scalar( 255, 0, 0 ), 4, 8, 0 );
}
}
VideoWriter Video("face_output.avi", CV_FOURCC('M','J','P','G'), fps, S, true);
if(!Video.isOpened())
{
printf("unable to write video file");
}
Video.write(frame);
//-- Show what you got
imshow( window_name, frame );
int c = waitKey(10);
if( (char)c == 'c' ) { break; }
}
return 0;
}
You are creating VideoWriter Video("face_output.avi", CV_FOURCC('M','J','P','G'), fps, S, true); inside the while(1), so you create a new Video in each iteration. Since you only write one frame per iteration, this is will be the only content of your face_output.avi file.
Try moving that line before the while(1):
// ...
Size S(dWidth,dHeight);
VideoWriter Video("face_output.avi", CV_FOURCC('M','J','P','G'), fps, S, true);
while(1)
{
// ...
I want to calculate optical flow on the left half and the right half of the captured video seperately. Using the results from the left and right half I want to eventually detect if the user(i.e. the camera) is moving forward or backward.
Currently, I have this system going for the left half of the screen. The code looks like this:
#include "opencv2/video/tracking.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include <iostream>
#include <ctype.h>
using namespace cv;
using namespace std;
static void help()
{
cout << "*** Using OpenCV version " << CV_VERSION <<" ***"<< endl;
cout << "\n\nUsage: \n"
"\tESC - quit the program\n"
"\tr - auto-initialize tracking\n"
"\tc - delete all the points\n"
"\tn - switch the \"night\" mode on/off\n"<< endl;
}
int main( int argc, char** argv )
{
help();
//Termination of the algo after 20 iterations or accuracy going under 0.03
TermCriteria termcrit(CV_TERMCRIT_ITER|CV_TERMCRIT_EPS, 20, 0.3);
Size subPixWinSize(10,10), winSize(31,31);
const int MAX_COUNT = 500;
bool needToInit = false;
bool nightMode = false;
//Video capture is from the default device i.e. the webcam
VideoCapture cap(0);
if( !cap.isOpened() )
{
cout << "Could not initialize capturing...\n";
return 0;
}
namedWindow( "Half screen Optical flow Demo!", 1 );
Mat gray, prevGray, image;
vector<Point2f> points[2];
for(;;)
{
Mat frame;
//Output from the Videocapture is piped to 'frame'
cap >> frame;
if( frame.empty() )
break;
frame.copyTo(image);
cvtColor(image, gray, COLOR_BGR2GRAY);
// Night mode not disabled
if( nightMode )
image = Scalar::all(0);
//This line saves into 'gray' only the left half of the original image. Thus making the detection of optical only on the left side possible.
gray = gray(Range(1,480), Range(1,320));
if( needToInit || points[0].size()<=5)
{
goodFeaturesToTrack(gray, points[1], MAX_COUNT, 0.01, 10, Mat(), 3, 0, 0.4);
cornerSubPix(gray, points[1], subPixWinSize, Size(-1,-1), termcrit);
}
else if( !points[0].empty() )
{
vector<uchar> status;
vector<float> err;
if(prevGray.empty())
gray.copyTo(prevGray);
calcOpticalFlowPyrLK(prevGray, gray, points[0], points[1], status, err, winSize, 3, termcrit, 0, 0.001);
size_t i, k;
for( i = k = 0; i < points[1].size(); i++ )
{
if( !status[i] )
continue;
points[1][k++] = points[1][i];
circle(image, points[1][i], 3, Scalar(0,255,0), -1, 8);
}
points[1].resize(k);
}
needToInit = false;
imshow("Half screen Optical flow Demo!", image);
char c = (char)waitKey(10);
if( c == 27 )
break;
switch( c )
{
case 'r':
needToInit = true;
break;
case 'c':
points[0].clear();
points[1].clear();
break;
case 'n':
nightMode = !nightMode;
break;
}
std::swap(points[1], points[0]);
cv::swap(prevGray, gray);
}
cap.release();
return 0;
}
How do I do this for the right half of the screen as well?
If I simply replicate the functions from above, say by modifying the image=image(Range(), Range()) values etc. I am not getting any output.
EDIT: Should I write out the part where the optical flow is calculated to a separate function and call it twice - once for the left half of the image and then for the right half?
I already had an optical flow code implemented using C++ in OpenCV. However, i would like to detect optical flow in half of the image frame. Which part should i edit? is it from this function below?
cvCalcOpticalFlowPyrLK(
frame1_1C, frame2_1C,
pyramid1, pyramid2,
frame1_features,
frame2_features,
number_of_features,
optical_flow_window,
5,
optical_flow_found_feature,
optical_flow_feature_error,
optical_flow_termination_criteria,
0 );
No. There are no changes necessary in the function itself. All you need to do is pass only the part of image on which you want to calculate optical flow to the function.
You can define the range of the image that you want to carry out the optical flow calculations on. using
wanted_image=image(Range(x1,y1), Range(x2,y2))
The following is a working code based on the lkdemo.cpp in the samples folder. THe only worthwhile change is
gray = gray(Range(1,480), Range(1,320));
//Gives the left half of the image
which defines the region of interest.
#include "opencv2/video/tracking.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include <iostream>
#include <ctype.h>
using namespace cv;
using namespace std;
static void help()
{
cout << "*** Using OpenCV version " << CV_VERSION <<" ***"<< endl;
cout << "\n\nUsage: \n"
"\tESC - quit the program\n"
"\tr - auto-initialize tracking\n"
"\tc - delete all the points\n"
"\tn - switch the \"night\" mode on/off\n"<< endl;
}
int main( int argc, char** argv )
{
help();
//Termination of the algo after 20 iterations or accuracy going under 0.03
TermCriteria termcrit(CV_TERMCRIT_ITER|CV_TERMCRIT_EPS, 20, 0.3);
Size subPixWinSize(10,10), winSize(31,31);
const int MAX_COUNT = 500;
bool needToInit = false;
bool nightMode = false;
//Video capture is from the default device i.e. the webcam
VideoCapture cap(0);
if( !cap.isOpened() )
{
cout << "Could not initialize capturing...\n";
return 0;
}
namedWindow( "Half screen Optical flow Demo!", 1 );
Mat gray, prevGray, image;
vector<Point2f> points[2];
for(;;)
{
Mat frame;
//Output from the Videocapture is piped to 'frame'
cap >> frame;
if( frame.empty() )
break;
frame.copyTo(image);
cvtColor(image, gray, COLOR_BGR2GRAY);
// Night mode not disabled
if( nightMode )
image = Scalar::all(0);
gray = gray(Range(1,480), Range(1,320));
if( needToInit || points[0].size()<=5)
{
goodFeaturesToTrack(gray, points[1], MAX_COUNT, 0.01, 10, Mat(), 3, 0, 0.4);
cornerSubPix(gray, points[1], subPixWinSize, Size(-1,-1), termcrit);
}
else if( !points[0].empty() )
{
vector<uchar> status;
vector<float> err;
if(prevGray.empty())
gray.copyTo(prevGray);
calcOpticalFlowPyrLK(prevGray, gray, points[0], points[1], status, err, winSize, 3, termcrit, 0, 0.001);
size_t i, k;
for( i = k = 0; i < points[1].size(); i++ )
{
if( !status[i] )
continue;
points[1][k++] = points[1][i];
circle(image, points[1][i], 3, Scalar(0,255,0), -1, 8);
}
points[1].resize(k);
}
needToInit = false;
imshow("Half screen Optical flow Demo!", image);
char c = (char)waitKey(10);
if( c == 27 )
break;
switch( c )
{
case 'r':
needToInit = true;
break;
case 'c':
points[0].clear();
points[1].clear();
break;
case 'n':
nightMode = !nightMode;
break;
}
std::swap(points[1], points[0]);
cv::swap(prevGray, gray);
}
cap.release();
return 0;
}
if you want to detect optical flow only in the half of the image, then you can simply give halves of the images (frame1_1C, frame2_1C) as parameters. For example, following code initializes a matrix belonging to the left half of frame1_1C:
cv::Mat frame1_1C_half(frame1_1C, cv::Range(0, frame1_1C.rows), cv::Range(0, frame1_1C.cols/2));