משתמש:Tamar Shalev1/multi aperture

In optics, Multi aperture represent a technology that shot multiple images for the same frame.

introduction

Structure of multi aperture camera

A single aperture constructed of a lens and a sensor. Multi aperture is a system that made of multiple number of lenses and sensors in a compact manner, in a symmetrical array. Pixels sizes remain the same, and each aperture acquire a complete image. The multi aperture has various abilities, for example: increase resolution, dynamic range and frame rate. Moreover, it assists decreases noises as RTS and dark current.

further information

Architecture of multi aperture imaging system

Comparison of (a) multi aperture camera and (b) single aperture counterpart

M pixel from every aperture are regarded as a sub-pixel, and they represent a large pixel whose area is M times larger than that of a single aperture counterpart Number of pixels for a duplicated image, in a multi aperture camera is as many as N*N, even though the number of pixels is actually M*N*N.

For a single aperture, an aberration can be fixed by adding more lenses to the camera. The extra lenses increase the weight of the camera significantly. In contrast, multi aperture has a synthetic f-number, this way it will allow high focus and light weight.

$f_{s}={f_{o} \over {\sqrt {m}}}$

• $f_{0}-f_{number}$of the original lenses
• m- number of pixels for a sub-pixel

In case the image is too dark or too light, or there are some blind spots, fabrication errors are made, and damage some pixels. In multi aperture a defective pixel can be removed without using interpolation, as it done in a single aperture.

In multi aperture, there are many pixels (N*N) for an identical objective point, and in case there is a defective pixel for a single aperture, the other apertures can calculate it.

Selective averaging

Variance calculation in multi aperture camera

In multi aperture imaging pictures system, N*N images are taken for a single picture simultaneously. Even though, there are different values for an identical objective point , because of various noise because of the pixel amplifier and the ADC which added to each of them.

$S_{m}^{2}={1 \over m^{2}}\sum _{i=1}^{m}\displaystyle \sigma _{i}^{2}$ $(1\leq m\leq N^{2})$

• the number of the selected apertures-m
• $\displaystyle \sigma _{i}^{2}$-variance of a single aperture
• $S_{m}^{2}$-combination variance

A variance is calculated for each aperture, the apertures who has the high variance will be excluded. Therefore, pixels who have high variance will be excluded automatically. The selectiveness of choosing apertures , gives it the name selective averaging. Noise like RTS and shot noise that caused by dark noise, create high variance. In this method, apertures who have high variance will be removed, so the noise will not influence.

SNR

Calculate SNR for a single aperture, influenced by sensor noise and shot noise $SNR_{SA}=20log_{10}{N_{e} \over {\sqrt {\displaystyle \sigma _{sensor}^{2}+N_{e}}}}$

• $N_{e}$-number of electrons in a signal
• $\displaystyle \sigma _{sensor}$-variance of the sensor

In multi aperture, the SNR is accepted by selective averaging in m apertures

$SNR_{MA}=20log_{10}{{\sqrt {m}}\cdot {N_{e}} \over {\sqrt {N_{e}+{\displaystyle \sigma }_{sensor}^{2}}}}$

Abilities

1. Facing different lightning- too light or too dark.
2. Accurate depth and distance.
3. Wide field of view.
4. Very high optical zoom.

Uses

1. Autonomous cars.
2. AR/VR/MR headsets & 360 cameras.
3. Survelliance & IoT cameras.
4. Action Cameras & Drones.

further reading

• • RTS Noise and Dark Current White Defects Reduction Using Selective Averaging Based on a Multi-Aperture System
    • Multi-Aperture-Based Probabilistic Noise Reduction of Random Telegraph Signal Noise and Photon Shot Noise in Semi-Photon-Counting Complementary-Metal-Oxide-Semiconductor Image Sensor