Capturing From 2 Cameras (OpenCV, Python)

Question

So I am attempting to capture from two cameras in openCV (python & windows 7). I capture from one camera just fine, youll also notice I am doing some funky stuff to the image but that doesn't matter. This is the code to attempt to use two

import cv
import time
cv.NamedWindow("camera", 1)
cv.NamedWindow("camera2", 1)
capture = cv.CaptureFromCAM(0)
capture2 = cv.CaptureFromCAM(1)
while True:
    img = cv.GetMat(cv.QueryFrame(capture))
    img2 = cv.GetMat(cv.QueryFrame(capture2))
    dst_image = cv.CloneMat(img)
    dst_image2 = cv.CloneMat(img2)
    cv.ConvertScale(img, dst_image, 255, -59745.0)
    cv.ConvertScale(img2, dst_image2, 255, -59745.0)
    cv.ShowImage("camera", dst_image)
    cv.ShowImage("camera2", dst_image2)
    if cv.WaitKey(10) == 27:
        cv.DestroyWindow("camera")
        cv.DestroyWindow("camera2")
        break

Rather simple. However it won't work. Upon trying to create the matrix from the second camera (second line of code in the loop), I am told that the capture is null. The cameras I am using are logitech and are the same model.

Side note: I also couldnt find the command to count cameras connected in python, so if someone could refer me to that I'd much appreciate it. --Ashley

EDIT: It might also be useful to know that windows often prompts me to choose which camera I would like to use. I can't seem to avoid this behavior. Additionally I downloaded some security like software that successful runs both cameras at once. It is not open source or anything like that. So clearly, this is possible.

Answer 1

I was having the same problem with two lifecam studio webcams. After a little reading, I think that problem related to overloading the bandwidth on the USB-bus. Both cameras began working if I 1.) lowered the resolution (320 x 240 each) or 2.) lowered the frame rate (~99 msec @ 800 x 600). Attached is the code that got I working:

import cv

cv.NamedWindow("Camera 1")
cv.NamedWindow("Camera 2")
video1 = cv.CaptureFromCAM(0)
cv.SetCaptureProperty(video1, cv.CV_CAP_PROP_FRAME_WIDTH, 800)
cv.SetCaptureProperty(video1, cv.CV_CAP_PROP_FRAME_HEIGHT, 600)

video2 = cv.CaptureFromCAM(1)
cv.SetCaptureProperty(video2, cv.CV_CAP_PROP_FRAME_WIDTH, 800)
cv.SetCaptureProperty(video2, cv.CV_CAP_PROP_FRAME_HEIGHT, 600)

loop = True
while(loop == True):
    frame1 = cv.QueryFrame(video1)
    frame2 = cv.QueryFrame(video2)
    cv.ShowImage("Camera 1", frame1)
    cv.ShowImage("Camera 2", frame2)
    char = cv.WaitKey(99)
    if (char == 27):
        loop = False

cv.DestroyWindow("Camera 1")
cv.DestroyWindow("Camera 2")

Answer 2

here is a small code:

import VideoCapture
cam0 = VideoCapture.Device(0)
cam1 = VideoCapture.Device(1)
im0 = cam0.getImage()
im1 = cam1.getImage()

im0 and im1 are PIL images. You can now use scipy to convert it into arrays as follows:

import scipy as sp
imarray0 = asarray(im0)
imarray1 = asarray(im1)

imarray0 and imarray1 are numpy 2D arrays, which you can furthere use with openCV functions.

Answer 3

In case you are using windows for coding, why dont you try VideoCapture module. It is very easy to use and gives a PIL image as output. You can later change it to a 2D array.