Subscriber Discussion

FOV Vs. Low Light Performance?

BE
Brian Elias
Aug 05, 2015

I have a 5MP (Axis P3367-VE) in an outside area that gets fairly dark at night. Right now I have the camera set at 50% of max FOV and 1/30s max shutter, and the images are "barely" usable* for facial capture. Unfortunately I don't have remote access and performing night-time tests are difficult. So, here's the question:

I only need to capture an area that's about 1/4 of the camera's maximum FOV. In terms of improving low light performance, is it better to keep the FOV as narrow as possible (greater pixel density, but less light enters the sensor) or as wide as possible (less pixel density, more light)? I couldn't find any reference to this in the IPVM testing.

Any help would be great. Thanks.

* barely usable meaning facial features are "distorted" due to camera operating at max gain; and lengthing the shutter time results in unacceptable motion blur

JH
John Honovich
Aug 05, 2015
IPVM

Narrower FoV should be better. Greater pixel density will help somewhat show more details though I am not sure if that will be enough.

I do not think the narrower FoV will have any issues with 'less light entering the sensor' unless that area is much darker than the rest of the existing FoV.

Btw, how about adding in an IR illuminator? Out of the question?

BE
Brian Elias
Aug 05, 2015

Unfortunately in this case the customer needs color images at night, so IR is out. And it's in a waterfront area with strict controls on area lighting... So, basically I need perfect 5MP color images in absolute darkness :-)

I may end up swapping the 1 5MP for 2 or 3 720p low-light cameras, but I'm trying to avoid this for aesthetic and cost reasons...

Thanks for quick reply though - much appreciated.

U
Undisclosed #1
Aug 12, 2015
IPVMU Certified

In terms of improving low light performance, is it better to keep the FOV as narrow as possible (greater pixel density, but less light enters the sensor) or as wide as possible (less pixel density, more light)?

Brian, this is an interesting question.

To start, whether the FOV is narrow or wide, we have the ALL the light there is from the ROI. In the wide shot we just have a lot of other light from the surrounding area, which is not needed. So less light overall in the narrow, because of the increased focal length, but ultimately the same amount of light coming from the area we are concerned with.

So my first thought would be to agree with John, and say that the narrow FOV is better because ALL the pixels in the sensor will be being used for, and only for, the ROI. This should increase the details captured in the scene. But since we are talking about a scene containing very low illumination, I'm not sure that's the 'final answer'.

Counterintuitively, and depending on how low the actual light levels are and the sensor's signal to noise ratio, a wider shot may give a better overall picture. The reason is because of the way that photodiodes work.

Each, pixel/photo diode is its own little accumulator of stored charge, created by the incident light from the scene striking it. But each photo diode also has its own individual accumulated noise component, created by ambient heat among other things.

Normally (when there is more than just minimal light), this noise is but a small percentage of the overall signal, and so isn't noticed. But when in extreme low light conditions, the percentage of noise rises to a noticeable level and can even exceed the 'real' signal.

So by using a narrow FOV you are actually spreading out all the light of the ROI among ALL the sensors pixels. So a given pixel might be 2 to 1, signal to noise. With a wide FOV, only a small part of the sensor is getting the light from the ROI. Therefore a given pixel might be 5 to 1 signal to noise.

So wide FOV gives less resolution but less noise (in the ROI) as well.

Did you get a chance to play around with the different focal lengths at all to see?

New discussion

Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.

Newest discussions