I'm working with kinect, and I need to save RAW depth image. This means that I shouldn't save it with a conversion to 8 bit (that is what imwrite is doing!) but save it as 16 bit, without have any bit-depth reducing. I hope that this question will not be too trivial, but I'm new to OpenCV programming. I tried the following, but it doesn't work:
[...]
Mat imageDepth ( 480, 640, CV_16UC1 );
Mat imageRGB;
// Video stream settings
VideoCapture capture;
capture.open( CAP_OPENNI );
if ( !capture.isOpened() ) {
cerr << "Cannot get video stream!" << endl;
exit ( EXIT_WITH_ERROR );
}
if ( !capture.grab() ) {
cerr << "Cannot grab images!" << endl;
exit ( EXIT_WITH_ERROR );
}
// Getting frames
if ( capture.retrieve( imageDepth, CAP_OPENNI_DISPARITY_MAP ) ) {
imwrite( fileDepth, imageDepth );
}
if( capture.retrieve( imageRGB, CAP_OPENNI_BGR_IMAGE ) ) {
imwrite( fileRGB, imageRGB );
}
return EXIT_WITH_SUCCESS;
Thanks in advance.
The problem wasn't in the way the image was saved, that was all right (if someone will have the same problem, be sure to save in PNG/TIFF format and specify CV_16UC1 when reading). It wasn't saved as 16bit because of VideoCapture; in fact I did the following:
if ( capture.retrieve( imageDepth, CAP_OPENNI_DISPARITY_MAP ) ) {
imwrite( fileDepth, imageDepth );
}
But the correct way to do it is:
if ( capture.retrieve( imageDepth, CAP_OPENNI_DEPTH_MAP ) ) {
imwrite( fileDepth, imageDepth );
}
So it was a silly problem.
Thanks to all the people who tried to help me.