int width = iplImage->width;
int height = iplImage->height;
uchar *iplImagePtr = (uchar *) iplImage->imageData;
uchar buf;
int limit = height * width;
for (int y = 0; y < limit; ++y) {
buf = iplImagePtr[2];
iplImagePtr[2] = iplImagePtr[0];
iplImagePtr[0] = buf;
iplImagePtr += 3;
}
QImage img((uchar *) iplImage->imageData, width, height,
QImage::Format_RGB888);
Refactorings
No refactoring yet !
Ants
April 2, 2010, April 02, 2010 09:10, permalink
No rating.
Login to rate!
Your code will only work given the following conditions:
iplImage->nChannels == 3
iplImage->depth == IPL_DEPTH_8U or IPL_DEPTH_8S
iplImage->imageData is 32-bit aligned (to satisfy QImage constructor requirements)
iplImage->widthStep % 4 == 0 (to satisfy QImage constructor requirements)
iplImage->width % 4 == 0 otherwise your for() loop will lose tracking with widthStep
Anyway, what's wrong with using cvConvertImage(src, dst, CV_CVTIMG_SWAP_RB)?
I am working with OpenCV and Qt, Opencv use BGR while Qt uses RGB , so I have to swap those 2 bytes for very big images.
Apart from just skipping two byte and hope nearby pixels have nearby values, there is a better way of doing the following?