Create a Virtual Camera to Test Scanning using Static Images or Videos
Creating a virtual camera is useful to test camera-dependent applications. We can use static images or videos as the image source for the virtual camera to test things like barcode reading, text recognition, and document scanning.
Adopting pyvirtualcam
Creating a virtual camera involves complex driver development. Fortunately, there have already been some projects like OBS for Windows/macOS and v4l2loopback for Linux, which can add a virtual camera to the system.
pyvirtualcam is a Python library that sends frames to the virtual camera based on OBS and v4l2loopback.
In this article, we are going to use pyvirtualcam to control the virtual camera.
Here is a video demonstrating the whole process:
Environment Setup
- Install Python3.
- Install Python packages:
pip install pyvirtualcam opencv-python Pillow numpy av -
Install the virtual camera:
For Windows and macOS, install OBS.
For Linux, install v4l2loopback following its instruction.
Using Videos
PyAV is a Pythonic binding for the FFmpeg libraries. We are going to use it to extract frames of videos and send them to the virtual camera.
Here is the complete code:
import av
import pyvirtualcam
import sys
def main(path):
container = av.open(path)
height = container.streams[0].codec_context.coded_height
width = container.streams[0].codec_context.coded_width
cam = pyvirtualcam.Camera(width=width, height=height, fps=20)
while True:
container = av.open("example.mp4")
stream = container.streams.video[0]
for frame in container.decode(stream):
frame = frame.to_ndarray(format='bgr24')
cam.send(frame)
cam.sleep_until_next_frame()
if __name__ == "__main__":
path = "example.mp4"
if len(sys.argv) == 2 :
path = sys.argv[1]
print(path)
main(path)
Using Images
We use OpenCV to read static images. Remember to convert the color space from BGR to RGB.
import pyvirtualcam
import cv2
import sys
import utils
def main(path):
frame = cv2.imread(path)
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
frame = utils.add_padding(frame)
width = frame.shape[1]
height = frame.shape[0]
cam = pyvirtualcam.Camera(width=width, height=height, fps=20)
while True:
cam.send(frame)
cam.sleep_until_next_frame()
if __name__ == "__main__":
path = "example.png"
if len(sys.argv) == 2 :
path = sys.argv[1]
print(path)
main(path)
Since the camera may have a resolution like 1280x720, the image will be scaled. We can add padding to it using the following code:
import cv2
def get_img_radio(img):
width = img.shape[1]
height = img.shape[0]
if width>height:
return width/height
else:
return height/width
def add_padding(img):
width = img.shape[1]
height = img.shape[0]
ratio = 16/9
desired_height = height
desired_width = width
top = 0
bottom = 0
left = 0
right = 0
if get_img_radio(img) > ratio: #17/9 > 16/9 add padding to short side
if width>height:
desired_width = width
desired_height = width / ratio
top = int((desired_height - height)/2)
bottom = top
else:
desired_width = height / ratio
desired_height = height
left = int((desired_width - width)/2)
right = left
else: # 4/3 < 16/9 add padding to long side
if width>=height:
desired_width = height * ratio
desired_height = height
left = int((desired_width - width)/2)
right = left
else:
desired_width = width
desired_height = width * ratio
top = int((desired_height - height)/2)
bottom = top
img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT,value=[255,255,255])
return img
Making the Image Draggable
We can make the image draggable to adjust its location in the camera stream.
-
Use OpenCV’s
imshowto show the image.def main(): width = img.shape[1] height = img.shape[0] cam = pyvirtualcam.Camera(width=width, height=height, fps=20) cv2.namedWindow("image", cv2.WINDOW_NORMAL) while True: time.sleep(0.05) img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) cam.send(img_rgb) cam.sleep_until_next_frame() cv2.imshow('image', img) k = cv2.waitKey(1) & 0xFF if k == 27: #esc break if __name__ == "__main__": path = "example.png" if len(sys.argv) == 2 : path = sys.argv[1] img = cv2.imread(path) img = utils.add_padding(img) vis = img.copy() main() -
Set up mouse callback to handle mouse events. If the left button is down, the image will be moved if the mouse moves.
ix,iy = -1,-1 previous_tx = 0 previous_ty = 0 current_tx = 0 current_ty = 0 is_mouse_down = False def handle_mouse_events(event,x,y,flags,param): global ix,iy, is_mouse_down, vis, img, cam if event == cv2.EVENT_LBUTTONDOWN: ix,iy = x,y is_mouse_down = True elif event == cv2.EVENT_MOUSEMOVE: tx = ix - x ty = iy - y if is_mouse_down: img = move_image(vis, -tx, -ty) elif event == cv2.EVENT_LBUTTONUP: is_mouse_down = False previous_tx = current_tx previous_ty = current_ty def main(): cv2.setMouseCallback('image', handle_mouse_events)warpAffineis used to shift the image.def move_image(image, tx,ty): global current_tx, current_ty tx = previous_tx + tx ty = previous_ty + ty current_tx = tx current_ty = ty rows, cols=image.shape[:2] moving_matrix=np.float64([[1,0,tx],[0,1,ty]]) return cv2.warpAffine(image, moving_matrix,(cols,rows))
Testing QR Code Scanning using the Virtual Camera
Let’s test the virtual camera using the JavaScript version of Dynamsoft Barcode Reader (DBR).
Here is the HTML file:
<!DOCTYPE html>
<html>
<body>
<script src="https://cdn.jsdelivr.net/npm/dynamsoft-javascript-barcode@9.0.0/dist/dbr.js"></script>
<script>
// initializes and uses the library
(async () => {
let scanner = await Dynamsoft.DBR.BarcodeScanner.createInstance();
let settings = await scanner.getRuntimeSettings();
/*
* The following code shrinks the decoding region by 25% on all sides
*/
settings.region.regionMeasuredByPercentage = 1;
settings.region.regionLeft = 25;
settings.region.regionTop = 25;
settings.region.regionRight = 75;
settings.region.regionBottom = 75;
await scanner.updateRuntimeSettings(settings);
scanner.onFrameRead = results => {
if (results.length > 0) console.log(results);
};
scanner.onUnduplicatedRead = (txt, result) => {
alert(txt);
};
await scanner.show();
})();
</script>
</body>
</html>
A scan region is set by updating DBR’s runtime settings.
Congratulations! You have now successfully created a virtual camera that can be used to test scanning using static images or videos.
Source Code
You can find the complete code here: