Grab the frame from gst pipeline to opencv with python

问题

I'm using OpenCV and GStreamer 0.10.

I use this pipeline to receive the MPEG ts packets over UDP with a custom socket sockfd provided by python and display it with xvimagesink, and it works perfectly. Following commend line is for this pipeline:

PIPELINE_DEF = "udpsrc do-timestamp=true name=src blocksize=1316 closefd=false buffer-size=5600 !" \
           "mpegtsdemux !" \
           "queue !" \
           "ffdec_h264 max-threads=0 !" \
           "ffmpegcolorspace !" \
           "xvimagesink name=video"

Now, I want to get one frame from this pipeline and display it with OpenCV. How can I do it? I know a lot about getting buffer data from appsink. But I still do not know how to convert those buffer to each frames for OpenCV. Thanks for reply, and any help :]

回答1:

Thanks, I have tried to use rtph264pay to broadcast the live video steam to udpsink. Following commend line is for the gst pipeline:

PIPELINE_DEF = 
"udpsrc name=src !" \               
"mpegtsdemux !" \       
"queue !" \               
"h264parse !" \ 
"rtph264pay !" \   
"udpsink host=127.0.0.1 port=5000"  

And I built a sdp file to make it can be received by opencv likes videocapture("123.sdp") 123.sdp, following content is for this sdp file:

c=IN IP4 127.0.0.1 
m=video 5000 RTP/AVP 96 
a=rtpmap:96 H264/90000 

It worked well now, just need to delete "blocksize=1316 closefd=false buffer-size=5600" to release the limitation.

来源：https://stackoverflow.com/questions/38546602/grab-the-frame-from-gst-pipeline-to-opencv-with-python