QImage is a Qt class for image handling (I/O and simple pixel access and manipulation), but it is far from a right choice for image processing. Interfacing Qt an OpenCV involves a natural need for converting QImage to OpenCV images of Mat. Here presents some solutions for their conversions.

1. From QImage to CV

   //Original

QImage lenaRGB=lena.convertToFormat(QImage::Format_RGB888);

ui->imgWindow->setPixmap(QPixmap::fromImage(lena,Qt::AutoColor));

intW=lenaRGB.width();

intH=lenaRGB.height();

//method1: trival

Mat lena_cv,out;

QImage lena2=lenaRGB.rgbSwapped(); 

lena_cv=Mat(H,W,CV_8UC3,lena2.bits(),lena2.bytesPerLine());

qDebug()<<"Imagesizebymethod1:"<<lena_cv.cols<<lena_cv.rows;

//cvtColor(lena_cv,out,CV_RGB2BGR);

namedWindow("Method1");

imshow("Method1",lena_cv);

//Method2: naive

Mat lena_cv2,lena_cv2_2;

lena_cv2.create(Size(W,H),CV_8UC3);

memcpy(lena_cv2.data,lena2.bits(),sizeof(char)*W*H*3);

namedWindow("Method2");

imshow("Method2",lena_cv2);

qDebug()<<"Imagechannels:"<<lena_cv2.channels();

//method3:robust 

Mat lena_cv3;

lena_cv3.create(Size(W,H),CV_8UC3);

unsignedchar*lena2_src=lena2.bits();

unsignedchar*lena2_dst=lena_cv3.data;

intstep=lena2.bytesPerLine();

for(introw=0;row<H;row++)

{

memcpy(lena2_dst+W*3*row,lena2_src+row*step,W*3);

}

namedWindow("Method3");

imshow("Method3",lena_cv3);

Comment:

1) swapping RGB in QImage to BGR in CV::Mat is needed prior to or after the converstion.

2. From CV to QImage

Sources:

http://qtandopencv.blogspot.com/2013/08/how-to-convert-between-cvmat-and-qimage.html

https://stackoverflow.com/questions/11886123/how-to-convert-qimage-to-opencv-mat

https://asmaloney.com/2013/11/code/converting-between-cvmat-and-qimage-or-qpixmap/