Oversampled Binary Image Sensor vs Active Pixel Sensor(CMOS)

Question

I am a beginner in the field of Image processing. I was designing a camera sensor when I came across these two type of sensors:

1. CMOS (Active Pixel Sensor).
2. Oversampled Binary Image Sensor.

Now, I can find many websites for comparing CMOS vs CCD but nowhere can I find a comparison with OBIS. Why is this? Is the OBIS too outdated to be used?

I was thinking of designing a camera that captures images using OBIS and later converts them into RGB format (with a variety of processes I have not discussed here). The advantages of Binary Sensor I mentioned Low image size and super fast processing. Are these really going to be an advantage over CMOS sensor? Will my camera be a success or is this a really bad idea trying to use Binary Image sensor? Is using Binary Sensor a good idea or is it a too outdated idea?

Thanks in advance for helping this novice!!

Answer 1

The success of your camera depends on too many factors! and they're out of the scope of this page. If the dynamic range requirements of your application are not too demanding then you could take advantage of the fast sampling rate and try to reconstruct a low quality image (by the same principle by which sigma delta modulation works, eg reconstructing a PCM signal from the output of a 1 bit output digital microphone). There's technique called halftoning, which has been the basis for reproducing images in print media since the invention of the newspaper. It is based on sampling an image with 1 bit and diffusing the error. Determining if using a Binary Sensor instead of a CMOS is a good or bad idea depends on your intended application. You should also pay attention to the size of the objects you want to picture. CMOS technology is widely used and you could probably find a cheap sensor to use in your camera.

Answer 2

At the time when the original question was asked, there was no hardware technology to physically construct an oversampled binary image sensor. Two years later - Technology is now slowly moving towards the construction of an array of millions (or billions) of sub-diffraction limit sized pixels that are required to realize such an oversampled binary sensor in practice. See: The Quanta Image Sensor: Every Photon Counts and Gigajot Technologies.

Different sensors suffer from different sources of noise. Do you care about SNR or dynamic range? Do you care about low light or high light levels? What is your target spectral responsivity i.e. quantum efficiency at different wavelengths? Finally, how much you care about cost and speed? Such questions make it very difficult to perform a fair apples-to-apples comparison between different sensor technologies.