kartikeygupta9097's profile - activity

Answering my own question here. Here are the steps I followed... Download OpenCV, MinGW, CMake. Install MinGW and CMa

TL;DR - I created my own blending procedure, but it's pretty basic...

Don't have enough karma points to reopen the question, but I figured out the solution. I imported the overlay image as a 4-channel image, and converted the underlay image into a 4-channel image. Then I set the origin of the overlay image on the underlay image, and ran a for loop(i=0->overlay.rows) within a for loop(j=0->overlay.cols). If the alpha parameter for the overlay[i][j] was 0, I set the underlay[i+origin.y][j+origin.x]'s RGBA to underlay[i][j]'s RGBA....

I have a transparent PNG image (shown below)...

and another image (shown below)

I want to copy the resize the first image and place it over the second image, and I tried doing so using the resize() and copyTo() functions. But the result was not what I expected or wanted...

How do I get rid of that green color in the background?

EDIT - So, I did further research and found that this is probably because I am using a three channel image and I need to use a four channel image. So how do I convert a 3-channel image to a 4-channel image?

Nvr mnd. I found out about the resize() func....

@berak So, is there a work-around to this or do I have to keep scaling down by 2 till I reach as close to the desired size?

The installer I downloaded from OpenCV's site had only a 'x64' folder. I use an i686 system. I want to build my libraries (don't know how) using GCC 6.2.0. If you could help me, it'd be cool of you. : D

Answering my own question here.

Here are the steps I followed...

1. Download OpenCV, MinGW, CMake. Install MinGW and CMake.
2. Run the OpenCV<version>.exe file. It'll ask you for the extraction location (I chose C:)
3. Then open cmake gui and choose the location "C:/opencv/sources" for source directory ,and choose whatever build (destination) directory you want. (I chose "C:/opencv-mingw")
4. Configure. Use default native compiler. Choose "MinGW Makefiles" as the compiler.
5. After hitting "Configure", I made sure that all the options under "BUILD_opencv_..." where checked. As @berak told me, if you want seperate, individual libraries, uncheck "BUILD_opencv_world" option.
6. Hit "Generate".
7. Open CMD, and navigate to the build (destination) folder.
8. Type mingw32-make or mingw-make, whatever does the job.
9. Let the building take its sweet time.
10. After the building process is complete, type mingw32-make install or mingw-make install, whatever your CMD understands.
11. Done.
12. You'll find your libraries at "<build-directory>/install/x86(or x64)/mingw/bin". Alternatively, you can also find them at "<build-directory>/bin" folder. Includes are present in the "<build-directory>/install/include" path. Set this path as the default include path.

P.S. - If you want to build the supplied example files as well, check the BUILD_EXAMPLES option in step 4.

So now, it displays the the output of the webcam for a quick second, and then the program terminates. What should I do to display it like a video, and just one peek at a still image.

Here's the python example that works fine... https://gist.github.com/kartikeygupta/15e10e02bb35e1fce821f8ff6ef723b9 (https://gist.github.com/kartikeygupta...)

It just displays one still image and that's it. The guy who wrote the code promised that it'll display a live, moving video stream.

@berak This is the link (https://gist.github.com/kartikeygupta/69b675d96829f52880dcb952c7a544c5 (https://gist.github.com/kartikeygupta...)).

Hi. I'm trying to capture video from my webcam via OpenCV C++. The program works fine, but the window terminates after displaying just one frame. I've tried Python examples as well on my computer, and they work as advertised. Help please.

#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <iostream>
using namespace cv;
using namespace std;

int main()
{
    VideoCapture stream1(0);   //0 is the id of video device.0 if you have only one camera.
    if (!stream1.isOpened())//check if video device has been initialised
        cout << "cannot open camera";
    //unconditional loop
    while (true)
    {
        Mat cameraFrame;
        stream1.read(cameraFrame);
        imshow("cam", cameraFrame);
        if (waitKey(0) >= 0)
            break;
    }
    return 0;
}

Didn't show up on google. I must have tried the wrong keywords.

Mat I;
//blah, blah, blah. Do Something....
if(I.isContinuous())
//do something.....

This a function from OpenCV's tutorial (Page Title - "How to scan images, lookup tables and time measurement with OpenCV" Subheading - "The Efficient Way")...

When it comes to performance you cannot beat the classic C style operator[] (pointer) access. Therefore, the most efficient method we can recommend for making the assignment is:

Mat& ScanImageAndReduceC(Mat& I, const uchar* const table)
{
// accept only char type matrices
CV_Assert(I.depth() == CV_8U);
int channels = I.channels();
int nRows = I.rows;
int nCols = I.cols * channels;
if (I.isContinuous())
{
    nCols *= nRows;
    nRows = 1;
}
int i,j;
uchar* p;
for( i = 0; i < nRows; ++i)
{
    p = I.ptr<uchar>(i);
    for ( j = 0; j < nCols; ++j)
    {
        p[j] = table[p[j]];
    }
}
return I;
}

No idea what happens here.

Thanks. Not sure what happened, but using opencv_world, instead of using the separate libraries, worked.

TL;DR - Certain code from OpenCV's own tutorial won't work on my system, when multiple examples already do...

I think I have correctly installed OpenCV 3.2.0 on my i686 system via MinGW 6.2.0, because when I was building my libraries, I also build that contours.cpp example, and it along with 3 other example files worked fine. I have now picked up this tutorial code from OpenCV's own Eclipse CDT documention.

#include <opencv2/opencv.hpp>
using namespace cv;
int main( int argc, char** argv )
{
  Mat image;
  image = imread( argv[1], 1 );
 if( argc != 2 || !image.data )
  {
      printf( "No image data \n" );
      return -1;
  }
 namedWindow( "Display Image", WINDOW_AUTOSIZE );
 imshow( "Display Image", image );
 waitKey(0);
 return 0;
}

but it displays the following error on my system...

undefined reference to `cv::imread(cv::String const&, int)'

Please tell me what I'm doing wrong...

You the man,@berak. Thanks.

@berak. So this time, I build the example files too, and their generated .exe files work just fine. So, can I assume that OpenCV has been successfully installed?.

I actually rage quit by deleting directory where I installed the files. So, right now, I'm rebuilding them. But I think I remember seeing a "opencv_world320.dll.a". So, in any of my further projects, Do I just link to that library, and that's it?

@berak So, does it mean that I can use OpenCV on my system?