After modifying one traditional color camera, this modified camera can capture depth of field (depth map) and color image in real time (not Kinect’s two cameras nor stereo vision). Kinect needs both depth camera and color camera; stereo vision needs two calibrated cameras, their devices are too expensive and large dimensions.

I use the LEDs as a lightning sourse because of the lower price. And Kinect uses laser that is too expensive and too hot
Kinect uses the technology “Laser Speckle” but our depth camera is not. So we can use LED for low price.

My friend, major electrical engineer, took one week by manual operation . He uses a soldering iron, a PCB, wire, some infrared leds and webcam module (connect to HP mini311). Then, I spent about six weeks for coding with OpenCV.

We use a traditional usb webcam module (CMOS) of developing board for this project. According to our theory, either COMS type or CCD type of image sensor is available.

To prove it is really a depth camera and also color camera, we use our DIY depth camera (only one camera) to automatically extract real foreground object from live view, and paste the object on another fake background film. Because this function, using Kinect, needs two cameras at the same time.

In the future, we hope that a lot of companies and low-price devices have this depth camera, not only for game.

Using 1 Camera to Integrate Live Foreground & Fake Background in Real Time

Uncategorized |