質問

I'm trying to capture images within node.js from a webcam connected to a Raspberry Pi. Capturing works fine, but now when I want to transmit the images I have some serious problems with framerate and lags.

Now my first idea was to convert the RGB images to 8bit grayscale, that should increase the performance by factor 3 (i hope..).

I'm using node.js and opencv-node for this. Here you can see some code snippet for the moment:

var startT = new Date().getTime();
var capture = new cv.VideoCapture();
var frame = new cv.Mat;
var grey = new cv.Mat;
var imgPath = __dirname + "/ramdisk/";
var frame_number = 0;

capture.open(0);

if (!capture.isOpened())
{
    console.log("aCapLive could not open capture!");
    return;
}

function ImgCap()
{
    var elapsed = (new Date().getTime() - startT) / 1000;

    capture.read(frame);
    cv.imwrite(imgPath+".bmp", frame);

    id = setImmediate(ImgCap);
}

ImgCap();

I tried to use something like

cv.cvtColor(frame, grey, "CV_BGR2GRAY");

after reading the image but i only get some TypeError saying that argument 2 has to be an integer... I dont know what to do at the moment. I referred to http://docs.opencv.org/doc/tutorials/introduction/load_save_image/load_save_image.html#explanation for converting a rgb to a grayscale image.

Beside I'm still not sure if this just gives mit a 24bit grayscale image instead of 8bit..?!

Thanks for any help in advance! :)

役に立ちましたか?

解決

use CV_BGR2GRAY instead of "CV_BGR2GRAY".

the former is an integral constant, the latter a char*

and, no fear, that will be 8bit grayscale.

ライセンス: CC-BY-SA帰属
所属していません StackOverflow
scroll top