class VideoCapture之VideoCapture::grab

Grabs the next frame from video file or capturing device.

C++: bool VideoCapture::grab()

Python: cv2.VideoCapture.grab() → retval

C: int cvGrabFrame(CvCapture* capture)

Python: cv.GrabFrame(capture) → int

The methods/functions grab the next frame from video file or camera and return true (non-zero) in the case of success.

The primary use of the function is in multi-camera environments, especially when the cameras do not have hardware synchronization. That is, you call VideoCapture::grab() for each camera and after that call the slower methodVideoCapture::retrieve() to decode and get frame from each camera. This way the overhead on demosaicing or motion jpeg decompression etc. is eliminated and the retrieved frames from different cameras will be closer in time.

Also, when a connected camera is multi-head (for example, a stereo camera or a Kinect device), the correct way of retrieving data from it is to call VideoCapture::grab first and then call VideoCapture::retrieve() one or more times with different values of the channel parameter.