Samsung, or rather Samsung’s Advanced Institute of Technology, has created what they claim is the first CMOS sensor that can collect both visible light data (which you’d use for a normal digital image) and depth data (like a Kinect). It’s accomplished by mixing in depth-sensing pixels with the RGB photosites normally found on such sensors. It was presented at ISSCC 2012 and reported by Tech-On.
The technology could be extremely influential: a small sensor that is able, with one lens, to determine the distance and size of objects it sees — the applications are extremely diverse. It could power autofocus, track gestures or individuals, or help determine the device’s position.
CMOS sensors are normally made up of a great number of light-sensitive photosites (or pixels) with filters on them to make them sensitive to only a certain range of wavelengths. Samsung has added a new type of pixel in there, four times as large as a normal one but able to detect depth data. The “Z” pixels (so the sensor could be called an RGBGZ sensor) use an established method for detecting distance by analyzing the differences between near-infrared light rays hitting the sensor.
The sensor, strictly speaking, doesn’t capture an RGB and depth image at the same time due to wavelength filter restrictions, but it can effectively time-share the available resources to make it appear as though that’s the case. It captures a
1280×720 1920×720 color image and a 480×360 depth image. Interestingly those two pixel counts are different aspect ratios, though why that is isn’t clear.
At the moment, the sensor is strictly a prototype and is unlikely to make its way to devices. However, improvements are already planned, for example a backside-illuminated type organization and the inevitable shrinking of the pixel pitch. The size of the sensor was not mentioned, but a few simple calculations suggest it is somewhere between 1/3″ and 1/2″ diagonally, perhaps around 5×3.5mm — about the size of a normal point-and-shoot sensor, perfectly capable of being put in a camera or phone.