When installing surveillance systems in low-light applications, security integrators must consider processing power, lens selection, IR-illumination technologies, camera positioning and other key factors.

Source: 13 Tips for Installing Surveillance Cameras in Low-Light Conditions – CE Pro

Jason Knott · November 21, 2017

When you used to install surveillance cameras in low-light conditions, you would blast the area with infrared light and hope to get some decent black-and-white video images. My how things have changed. New lenses, processors, lighting options and video analytics can help security-conscious users capture clear, full-color images in poorly lit areas.

We talked to video-surveillance experts to provide professional security dealers and integrators with tips to optimize video capture in low-light conditions. They advised on everything from selling your services to positioning cameras to implementing video analytics.

1. Use the Right Terminology to Talk the Talk

Before even beginning to spec or buy the right low-light camera, dealers can find themselves confused by a slew of labels being used to reference this category. Among them:

  • Low-Light
  • Day/Night
  • Starlight
  • SenseUp
  • Night Vision
  • Light Finder
  • Light Toucher
  • Dark Finder
  • Light Catcher
  • Thermal Imaging

So which is the proper terminology? All of them! Those monikers refer to the same classification of cameras. Many are trademark wording that camera manufacturers have placed on their technology. Among the generic, nontrademarked terms are “low-light” or “day/night.”

“It can get somewhat confusing,” admits Joel White, regional marketing manager for Bosch Security Systems.

2. Understanding IR Illumination

Even though low-light cameras have varying labels, they contain the same fundamental components: a lens and sensor along with some level of image processing. These are not be confused with thermal cameras or cameras with IR illuminators, which are quite different.

The most common low-light cameras use an IR cut filter, which accounts for 90% of the market, according to White.

An IR cut filter is a mechanical filter that sits between the lens and the sensor (CMOS chip) and cuts out IR illumination during the day to improve color quality. It is called a cut filter because it filters out, or “cuts out” IR during the day. At night it slides out of the way to allow more light to get to the sensor, which improves low-light video quality.

SnapAV’s 4MP Luma 500 outdoor cameras include day/night mode and smart motion sensing.

When the IR cut filter is out of the imaging path, the image is monochrome without color. In most cases, the filter is mechanically driven by algorithm, but in some cameras, it can be manually controlled.

But if almost all cameras on the market contain IR cut filters, what distinguishes one from the next? The lens and the processing.

The lens transmits light to the sensor, and the data on the sensor is encoded and processed by a processor. The variance among cameras is often in the optics (lens).

According to Ryan Zatolokin, senior technologist at Axis Communications, it is important to use an IR-corrected lens, which costs more.

“If you do not use an IR-correct lens, then the flow of the light comes in a slightly different angle, which causes the image to look blurry even in a static image because the IR light is hitting the image sensor differently than the visible light,” he notes.

White adds, “If you have cheesy optics in front of the IR filter, you are not going to get much quality out of that camera and will have wasted the money on the sensor.”

There is a wide variety of sensor capability. Most premium-branded manufacturers of low-light products use the highest quality sensor they can. Many imaging sensors are supplied by the same manufacturers, including SonyPanasonic and Aptina (now ON Semiconductor), and bought by various camera manufacturers, in some cases to address certain applications.

According to Robert Mitchell, director government practice and law enforcement for IC Realtime, Sony’s new Starvis sensor technology, for example, is designed for industrial applications.

In other improvements, many cameras now manipulate the light by doing things like increasing the time at which the lens is open so it can absorb more light. This can even be done to the point of allowing the camera to retain a color image at night, which has both benefits and limitations.

“When you have the aperture open longer, you get this time-lapse effect where any movement would be blurred, and you see a trailing image.”

— Alex Petrao, SnapAV

Keeping the aperture open for too long, however, might give you nice color pictures, but the practice has negative impacts.

“When you have the aperture open longer, you get this time-lapse effect where any movement would be blurred, and you see a trailing image,” says Alex Petrao, director of products for surveillance at SnapAV. “Now, you’re starting to see some significant improvements in the sensors being used, so that they’re able to just pick up very low amounts of light and still be able to retain a color image.”

None of this should be confused with thermal cameras, another type of technology that works well in low light because it looks for heat, not motion or images.

Market leaders like FLIR offer thermal cameras that can detect people from a long distance, even up to 1 kilometer away. These thermal cameras are also finding new applications for the smart home and office. FLIR and other thermal cameras can detect heat loss in a home, and are so sensitive as to be able to detect when a baby has a fever.

3. Processing Power, Image Toning are Key

So if companies are buying sensors of the same size and resolution from the same supplier, how can one product be superior or inferior? That’s where processing comes into play.

All things being equal with two cameras using the exact same sensor (and the same lens), the proprietary processing technology is where all the “magic” takes place. Having said that, many manufacturers also employ the same OEM processor, from companies like Amberella or Texas Instruments, and then tweak them.

The control of tuning the image is key. You can tune an image during daylight to perfection, but find that the same level of tuning at night under a sodium vapor light, or in complete darkness, does not work at all. Tuning an image is a very exact and time-consuming science.

Some companies just use the default image path in the processor. A company that does not have depth in its engineering staff will take those off-the-shelf components and deliver it. The camera will work fine but it won’t be better. Image toning, noise suppression, and the ability to maintain color and contrast in low light is what differentiates one camera manufacturer from another.

4. ‘Low-Light May Take on Different Meanings’

In a typical parking lot with minimal lighting, you might have 5 lux of light. That isthe  equivalent of walking into your closet and shutting the door. But we now have cameras that can maintain a color image in light conditions where you could not even see your hand in front of your face.

A typical IR camera will capture images between 1 lux and 0.1 lux. Some of these starlight-type cameras will start around .01 lux and go down to 0.00001 lux. “I have seen those cameras get some pretty impressive images where it seems almost completely dark and produces a view that’s nearly daytime-looking video,” says SnapAV’s Petrao.

According to Mitchell at IC Realtime, Sony’s Starvis sensors can increase light sensitivity dramatically. Indeed, new low- and ultralow-light sensors are now able to pick up an image at incredibly low light levels; however, camera makers are starting to get caught up in the marketing game,racing to see how many zeros they can put in front of their lux light levels. (Can you say, “0.00001 lux”?)

“You’re starting to get these improvements in the sensor technology,” says Petrao. “Each individual pixel is getting better at picking up light. These are very low-light sensors that can pick up an image that would have been a completely black image just a few years ago.”

He continues, “They can actually make an almost black picture look like a near-daytime picture. There is some pretty impressive stuff going on. Most of that is done through the sensor,” he says. “And there’s some differences between camera manufacturers because on top of the sensor they do some various amounts of processing to clean up the image and make the color better. These are some of the technology advancements you’re seeing, and it’s moving pretty quickly over the next couple years.”

The reality is that very rarely will a security integrator encounter a scenario where there is complete darkness. There is likely always going to be some ambient light to work with from street lamps, the moon or even the stars.

5. Beware of Spec Sheet Claims

As noted, some manufacturers will make astronomical claims about their cameras’ low-light sensitivity and put it in their spec sheets.

“A lot of them are simply reporting sensor specifications as opposed to true camera performance. A sensor spec is only going to tell you the minimal light level in which the sensor can obtain some sort of image definition, but realistically, a camera performing with any sort of reliable image at 0.00001 lux is near impossible,” explains Bosch’s White.

The key criterion for integrators is to isolate on finding a camera that can deliver surveillance-quality video under the minimal illumination situations.

Bosch, for example, has demos set up at its facility along with comparisons. Most good integrators will want samples they can bench test on their own. They can also look at reviews from independent third parties.

Bosch says its Starlight technology enables low-light cameras to capture colors in dark environments, while competitors’ products can only pick up black-and-white. The company recommends bench-testing cameras or seeking demos before selecting surveillance product lines.

6. Resolution May Not Impact Results Like You Think

A high megapixel camera does not always equate to a better low-light image. It might be counterintuitive, but consider the number crunching, SnapAV’s Petrao explains.

“As you get higher resolution, and higher megapixels, essentially what ends up happening is each pixel becomes a smaller percentage of that sensor,” he notes. “Think of the sensor as a CMOS chip that sits on a board in the camera. That chip tends to be a certain size, maybe 1½ inches. If you double the resolution in that fixed size, you’re basically making the sensor for each pixel smaller.”

Petrao says that as the sensor for each pixel gets smaller, there’s a need to increase the amount of sensitivity of that sensor to maintain the same level of quality.

“It’s an area game,” he says.

7. Check Lens Speed

The speed of the lens is important. A datasheet will always quote the light level at a certain lens speed. But dealers need to be careful because, for instance, sometimes a spec sheet will indicate a light level at f1.4 but the camera itself can only accommodate a lens at f2.

A fast lens is important and spending more money on the optics is critical, camera manufacturers say. Depending on how an integrator chooses to carve up his bill of materials, spending on the lens might be the most expensive part of their camera expenditures. And if you put a slow lens on a camera, it doesn’t matter’s behind it because the lens is going to determine what information reaches the sensor.

8. Proper Positioning Is Important

Cameras will always be deployed to cover the field of view based on the need. Especially in an outdoor application, it is really hard not to have some sort of light source in the scene. But there can be bright lights in the field of view as long as they are not directionally pointed right at the camera.

“The positioning of cameras, in general, is very important,” says Petrao. “It’s extremely important how you position cameras to make sure you pick up what you want to pick up. Often, integrators put cameras where they can run a wire, rather than really focusing on what are they trying to capture, and what level of detail they want to capture.”

Positioning the camera so that the common range of motion is moving across the field of view is more effective than positioning the camera so the common movement is coming toward the camera.

Avoiding bright light pointing directly at the lens is another good idea. That can cause flare or what some people may see as “fog” on the image.

Thermal cameras from companies like FLIR can detect people from 1 kilometer away, but not for identification purposes.

The field of view in terms of distance is also vital. The level of detail from the camera is highly dictated by how close the camera is, and how much it’s zoomed in.

“Often, integrators will just put a camera and set it to its widest angle to pick up as much as possible, and then the customer complains when they can’t tell who that person is walking across their yard,” says Petrao. “Yeah, it’s because that person is really far away from the camera and the camera is not zoomed in.”

9. Factor Distance into Equation

You need to have the right combination of lens and camera that takes into consideration the distance from the area you are trying to monitor when you want to use analytics with the camera.

“I’ve talked with customers who want to put a camera 250 yards away at the widest setting, and then they want to detect motion. It’s just not going to happen,” says White. “The only way to detect at that distance effectively is to move over to thermal, which has a big tradeoff because the maximum resolution for thermal is VGA. Also, thermal has no color. It is not even monochrome … it almost looks like a negative. Thermal cameras can detect at a very long distance. You will certainly detect activity and you might be able to determine if it is a person or an animal.”

About 80% to 85% of cameras sold in North America are domes so you are not going to have many options to change the lens. Once you move into a box-style camera, you have lots of options.

10. Enlist Video Analytics

The ability to use video analytics in a low-light camera is a pretty black-and-white issue. As long as you have a usable image, you can do it — there are no limitations on analytics, including motion detection. The only limitation is if you have a poor image, analytics will not work well.

Oddly enough, sometimes having super high resolution can work against you. According to Zatolokin from Axis, megapixel resolution can hurt low-light performance.

“It is simple physics,” he says. “You have more pixels on the sensor and you have to hit each pixel with the same amount of photons. But the good news is that new technology is compensating for this. Sensors are becoming more light sensitive as time goes on.”

This scenario can come into play when considering the use of a 360-degree camera with multiple high-resolution sensors for a low-light application.

“Just because you are using multiple sensors or a wide angle technology does not necessarily mean you cannot use day/night,” adds Zatolokin.

11. Don’t Expect to Use Facial Recognition

Speaking of analytics, facial recognition is common application. It can be done in low-light scenarios, but requires a certain amount of pixel resolution, usually between the person’s eyes. So the camera has to be close enough and have enough light. The proximity can be achieved with a zoom lens if necessary, so conceivably it can be done.

“It is not facial recognition per se, but last month at Bosch HQ there was a big snowstorm and a bunch of local kids came out after 10 p.m. to do some sledding in the parking lot. We have a nine-camera demo system set up covering the parking lot, and one of our employees recognized his son.”

IC Realtime says its EL-400PIXIM incorporates WDR and a three-layer IR mechanical filture to provide detailed images in “the most difficult lighting conditions.”

12. IR Illumination and Imaging

Like any surveillance camera project, the equipment has to fit the application. In an environment where there is some level of illumination, finding a camera that will give you a solid usable image should not be a problem. But what if there is no light at all? In that case, dealers might consider a camera with IR LEDs built in to it, or even a thermal camera, which requires no light at all.

In a low-light application where there is a “hot spot” from a parking lot lamp, for example, cameras with Wide Dynamic Range (WDR) are able to combine the best portions of multiple frames to create a clearer image.

“It is all done super-fast at 30fps so you are not introducing any motion blur in to the picture. WDR also helps you see a subject that is very close to the camera as well as someone in the background who is far away. Technologies can help compensate for that [type of application],” says Axis’ Zatolokin.

Products from companies like Illuminar provide IR and white light illuminators that feature the latest high-efficiency Osram LEDs that consume a minimal amount of power. The illuminators are enclosed in an IP67 weatherproof housing, so they can endure harsh weather environments. These lighting products can be selected based on the field of view needed for the application, from panoramic to narrow. Indeed, Illuminar’s IR919-POE-2 Series can illuminate up to 919 feet away. The company partners with various manufacturers to provide illuminators for their cameras.

13. Rule of Thumb: Buy What You Know

Because there is a race to zero taking place in the marketing world in terms of the lux levels of the cameras, White says it is vital security dealers are comfortable with their vendor.

“Because there are no standards in this industry to measure low-light capability, it really becomes a question of how do you as a customer know what you are buying,” he advises.

That means narrowing it down to a few companies that fit the application and the price range. He says it behooves an integrator to put cameras through their paces, either on a jobsite or in a lab test, so they can see the differences and make the best decisions.