Change videoCapture format to MJPEG

asked 2018-03-17 03:15:58 -0600

quentinboul
1 ●1 ●1 ●1

Hi there,

I would like to get frames from my camera in the MJPEG format. I know for sure that my camera is able to use the MJPEG but when I try:

webcam.set(CV_CAP_PROP_FOURCC, CV_FOURCC('M', 'J', 'P', 'G'));

I get:

HIGHGUI ERROR: V4L: Property <unknown property="" string="">(6) not supported by device

Can you help?

Thanks!

Here is my code:

#include "opencv2/opencv.hpp"
#include <iostream>
#include <sstream>

using namespace std;
using namespace cv;

int main() {
    VideoCapture webcam;
    webcam.open(0);
    webcam.set(CV_CAP_PROP_FOURCC, CV_FOURCC('M', 'J', 'P', 'G'));
    webcam.set(CV_CAP_PROP_FRAME_WIDTH, 1920);
    webcam.set(CV_CAP_PROP_FRAME_HEIGHT, 1080);

    Mat frame;
    webcam >> frame;         //get a new frame from camera
    imwrite("image.jpg", frame);
    return 0;
}

edit retag flag offensive close merge delete

Comments

it is not possible, to change the fourcc of a webcam, using VideoCapture.

why do you want that ?

the next thing you'd have to do would be: uncompress it again, else you can't use it in the code above.

you can try one of those , or to disable CAP_PROP_CONVERT_RGB (which will result in some kind of native YUV)

berak ( 2018-03-17 03:23:38 -0600 )edit

I have a See3Cam from e-con system. (https://www.e-consystems.com/4k-usb-c...) I really need my camera to take the picture as fast as possible and they say on their datasheet that the frame rate is higher when using MJPEG than UYVY . that's why I want to use MJPEG. But I don't know how to do that.

quentinboul ( 2018-03-17 04:11:13 -0600 )edit

libv4l2 (or at least, opencv's wrapper) does not support it
getting images over the wire fast is one thing, but again, you'd need another (expensive) imdecode() call on the other side, to use the image

berak ( 2018-03-17 04:17:31 -0600 )edit

Do you mean that the time my camera would take to capture the frame wouldn't be different if I use MJPEG intead of UYVY? I don't really care about the time to get the images over the wire and the time my SoM takes to analyse the picture but I want the time to capture the frame to be as fast as possible. Thank you for your answers!

quentinboul ( 2018-03-17 04:39:38 -0600 )edit

"the time to capture the frame to be as fast as possible" -- that's too broad. (and somewhat naive)

there's processing on the webcam (protocoll / compression)
then, there's transport over usb
then, there's cap_libv4l2, processing there, too (convert whatever we got to bgr, or not ?)
then, there's your app, what are you doing, what do you need ? (e.g., if all you need is grayscale, converting to/from bgr would be a waste, and you'd better just extract Y from YUV)

i guess, in the end, you have to profile / analyze all the steps involved, and take the optimal one for your application

P.S. did you consider to use libv4l directly, shortcutting the VideoCapture ?

berak ( 2018-03-17 04:50:19 -0600 )edit

add a comment

Change videoCapture format to MJPEG

Comments

2 answers

Links

Question Tools

Stats

Related questions

Change videoCapture format to MJPEG edit

Comments

2 answers

Links

Question Tools

Stats

Related questions

Change videoCapture format to MJPEG