3DV systems : ZCam - depth camera

ZCam is the RGB and depth cameras for gesture application. 3DV systems joined Microsoft Natal project for gaming technology development, and later they were acquired by Microsoft in 2009.

ZCam
You can refer to the description on wiki (http://en.wikipedia.org/wiki/ZCam). The apparatus consist of RGB camera, depth camera, and IR light source, but here we focus on "depth" camera and technology.

Spec

• Depth sensor format type: VGA,QVGA,QQVGA
• Operating range: 0.5~2.5m
• Range resolution: 8 bit
• Resolvable depth: 1~2cm
• Framerate: up to 60fps

 For extracting depth information, 3DV uses IR light emitter, detector, and "modulator". Light source emits a pulses train, and detector receives the light reflected from object. Based on time-of-flight concept, ZCam is a type of "gate camera" so a modulation technique is requires.

Principles
Gate camera is derivative of time-of-flight technology which calculates the distance from the time interval between emitter and receiver that they both have modulation to form a time signal.But camera use area image sensor and can't operate such high frequency and short exposure time, especially for short distance range - light pulse reflect very soon, and so that's why detector needs a modulator.
As IR pulse emit at t0, pulse will reflect to detector at t . Modulator would open its gate to let light pass at tgs and close at tge, and detector will integral the light in the open gate duration. Different position has different time delay to detector, so the difference between delay and gate (tgs) will cause the integral intensity variation. But there is still a important issue in this technique - how do you know the intensity variation is caused from distance, not from reflectivity? It still needs a normalization - use a longer duration gate to integral the complete reflected pulse to get the base intensity, and then divide the intensity of short gate.

For simplicity, light pulse and gate has the same time duration.The integral function between light pulse and modulation gate is a triangle function - It reach max as gate has complete overlap with pulse. This is a ideal case and don't forget the raise/fall time and duration deviation. More, it then needs to consider the arrangement the sequence of modulation to avoid ambiguity problem.

Modulator
Light source and detector both have modulation, but light source can be controlled directly (ex: LED, laser diode, VCSEL). The most important issue for the gate technique is the modulator for detector. The modulator must operate very fast (<1ns is required)and synchronize with controller.3DV systems had involved in this field for a long time, and we can see their progress from their patents disclosed.
Image intensifier à Solid state optical shutter à gate pixel

• Image intensifier - like a night vision camera. There is a microchannel to intensify photoelectron.

• Solid state optical shutter - a deviation from image intensifier, but there is no microchannel. The light will be absorbed in PD layer, and photoelectrons pass the intermediate substrate and then LED layer reproduce light. 

• Gate pixel - Detector has a built-in switch in each pixel, so it doesn't need modulator in front of camera.

Depth camera
Depth camera is very hot recently, especially after Microsoft Kinect debuts. Microsoft first acquired 3DV systems and get the gate camera technique and patents, and more extended to time-of-flight technique after acquiring Canesta. Gate camera is good, but PrimeSense's structured light technique overshadow 3DV systems. Each of these 3 companies has different solution to get depth information. Maybe it is that a modulator needed at that time makes 3DV less competitive.

By the way, 3DV and Microsoft deploy some patents about the connection between color camera and depth camera located in parallel - like how to adjust the two image, how to extract object...etc.