How to Handle Headlight Surveillance Problems

By: John Honovich, Published on May 17, 2011

Headlights are a painful problem for video surveillance. Figuring out how to overcome headlight's glare is both a fascinating and frustrating endeavor. Unfortunately, these challenges have not received enough attention and focus. If you want to capture license plates, regardless of how much resolution you have, you are going to have to deal with headlights.

Below are 2 image samples from a recent intersection test that show how bad surveillance quality can get when dealing with headlights:

That's pretty disastrous and maybe a worse case scenario because there are a lot of cars. However, even with a single car, the image below shows the problems you often face:

We have been doing numerous tests and conducting research into best practices and options. This report is a pre-requisite and companion to our License Plate Capture Shootout results. However, in performing these tests, we realized that the impact of headlights is so severe and so central that it demands a dedicated report explaining the issues involved and providing recommendations on how to overcome.

Inside the Pro report, we look at the following:

  • Why Dealing with Headlights is So Hard
  • Understanding Variations and Issues Amongst Headlamps
  • Best Ways to Adjust Exposure
  • Choosing Color vs Black & White
  • The Limits of Megapixel
  • The Limits of Using IR and White Light
  • Using Filters and Understanding Tradeoffs

Recommendations and Setting Expectations

The most important expectation to set with headlights is choosing between these mutually exclusive scenarios:

  • Almost always capture license plates clearly but almost never capture general scene activity (like someone walking across the street)
  • Sometimes capture license plates clearly while capturing general scene activity

While we see significant interest in doing both, practically speaking, you need to choose between one or the other. The only way to do both is to dedicate two cameras to the location.

That noted, we do believe some optimizations can reduce headlight problems, allowing the capture of some license plates while monitoring the general scene. Inside we explain the specifics but here are the key recommendations for using regular surveillance cameras:

  • You will not capture all the plates. Even with using multiple 'tweaks' you might still miss half of the plates.
  • The problem with headlights is a strong WDR issue - not feasible to overcome without tradeoffs.
  • Headlights vary significantly so an optimization for one vehicle's headlights can easily make it worse for another vehicles.
  • Shortening the exposure will reduce the glare from headlights but will also make the overall scene darker - often an unacceptable tradeoff for users.
  • Color might be better than using black & white as it tends to produce less headlight glare.
  • Higher resolution can help provide details but will not solve the key problem of headlight glare.
  • Adding in illuminators (IR or white light) can help moderately.
  • Adding on IR bandpass filters to a regular surveillance camera is not recommended as it is costly, has significant tradeoffs and does not substantially improve license plate capture.

Why Dealing with Headlights is So Hard

Get Notified of Video Surveillance Breaking News
Get Notified of Video Surveillance Breaking News

Dealing with headlights combines the worst aspects of WDR and low light scenes. Individually, these are two of the biggest problems in surveillance. Combined, it's the 'perfect storm'. Let's break them down:

In WDR, cameras are challenged in handling the wide variations in light within the scene. Here's an example scene:

Over a few feet, the lighting drops significantly (a variation of about 6.5x). The result is that the image quality suffers - usually either the darker or the lighter part as the camera needs to 'pick' one to optimize.

Typical WDR surveillance scenes occur during the daytime so absolute low light is not a problem. However, when dealing with headlights at night, this does become an issue.

Here's an example of a vehicle with their headlights turned off:

Even at approximately 5 lux, the visible noise is clearly impacting the clarity of the image and, in particular, the legibility of the license plate. This is a fundamental aspect of night time surveillance. Gain is necessary but reduces the overall resolving power of the image. The only other option is to lengthen exposure. While that will allow a reduction in gain levels, it will result in severe motion blurring of moving objects.

Putting this together, let's analyze the challenges facing surveillance of a vehicle with its headlights on:

Headlight scenes combine both challenges - low light and WDR. Indeed, the WDR challenge for headlights is actually worse than what is commonly seen in indoor / outdoor doorways - the most common reference case for WDR challenges. While the specific lux levels of headlights vary based on how and where you measure the light, the variance between the area in front of the headlights and the license plates (just a few feet away) is always very steep.

Understanding Variations and Issues Amongst Headlamps

What makes dealing with headlights even more challenging are the significant variations that exist among headlamps. Here are key elements to keep in mind:

  • The light output / strength of headlights can vary significantly. While overall regulations exist, the output power can vary and the level of grime, dirt or other obstructions can impact the overall strength of the beam.
  • Where the headlight is aimed can vary. While standards exist and inspections occur in many regions, the aim of the headlight can shift over time - impacting where the light is projected and how close it is to the license plate.
  • The style of the headlamp: Some headlamps direct the light more towards the outer side, others in a circular fashion straight ahead and still some more towards the center. Even with the same level of light output, the impact of the headlight on a license plate can vary notably.
  • The distance between the headlamp and the license plate. All things being equal the closer a headlamp is to a license plate, the worse the lighting challenges will be. Some vehicles have their headlight a foot or two above the plate, others have them on the same vertical level as the plate.

The level of variations and issues with headlights is pretty interesting. The amount of information online is fairly weak with the best primer on headlamps / headlights being the Wikipedia entry (which is well written and thorough).

Because of these headlamp variations, a tweak or trick that improves surveillance quality for some vehicles is unlikely to work for all vehicles. This is important to keep in mind as one attempts to optimize camera settings.

Best Ways to Adjust Exposure

When looking at optimizations, the most common one recommended is to to shorten the exposure time.

Here's an example of an ideal outcome of shortening the exposure time:

Keep in mind this is with a white light LED illuminator turned on. The appropriate exposures would be significantly different without the added illumination

Moreover, important limitations exist in optimizing the shutter speed:

  • There is no single 'right' exposure setting. People will often talk about 1/60s, 1/120s, 1/250s or 1/1000s exposure settings. It depends on the overall brightness of the scene one is optimizing.
  • There is no single exposure setting that will work with all vehicles using 'regular' surveillance cameras. Because of variations in headlamp lighting, one exposure setting may be too dark for vehicle A but just right for vehicle B. If the exposure setting is slowed to provide more light for vehicle A, vehicle B's plate is likely to be over-exposed.
  • Shortening the exposure time will darken the rest of the scene and can make the overall camera unusable for anything but license plates. This often runs counter to expectations of users who, understandably, would like their cameras to see both people clearly in the background as well as the characters of a license plate. Unfortunately, this is generally not feasible.

Choosing Color vs Black & White

Normally, when doing night time surveillance, the go to choice is black & white (using Day / Night cameras). Surprisingly, we have seen fairly consistently that black & white modes produce significant more headlight glare than color modes. Here is an example of the same camera, same scene contrasting B&W vs Color:

The Limits of Megapixel

While higher resolution can help, it does not address the main problems in handling headlights - glaring lights. If you can overcome or mitigate the headlights, higher resolution will help provide greater details and/or expand the coverage area. However, most megapixel cameras use slow shutter defaults. These defaults regularly wash out the front of the vehicle, especially if the vehicle is moving. Pay careful attention to shortening the shutter speed.

Here's an example of SD vs a 3MP camera. The 3MP camera shows more details but the headlights increase glare significantly:

The Limits of Using IR and White Light

The second most common recommendation after shortening the shutter is to add IR or white light illuminators. In our tests, these can help moderately but will not completely overcome the power of headlights. We are assuming users will add surveillance 'level' spot illuminators. These will increase the effective illumination of the scene by 5 - 20 lux (depending on the strength of the illuminators and the distance of the target from the illuminator). This is sufficient to increase the probability of capturing license plates but might make some plates overexposed (similar to how changes in the exposure setting will make some plates over and others underexposed).

Here is a simple contrast for 3 lighting types for a single car and single camera. The key point is that the performance variances are modest. It is incorrect to assume that white light LED is better than IR as these results will vary based on the camera used and the license plate. The goal is to primarily show how similar the scene remains for all 3 lighting options:

Using Filters and Understanding Tradeoffs

The last element we will review is using IR bandpass filters. These are very common in dedicated or specialized license plate capture cameras. What these filters do is block all visible light and only allow IR light (e.g., 850nm, 940nm, etc.).

If you are looking for a camera dedicated to capturing license plates and are willing to forgo capturing the overall scene, we recommend buying such a camera with such a filter built in.

While you can buy an add on IR filter and use on a 'regular' surveillance camera, we do not recommend it:

  • Your camera can now only operate in black and white mode 24/7. These add-on filters are mounted in front of the lens, blocking out color 24/7.
  • Headlights generate high levels of IR light. In our tests of an add-on IR bandpass filter, the filter only removes a slight level of headlight glare. The camera still picks up a very high level of IR light from the headlight. As such, the performance improvement is not great.
  • Surveillance lenses generally do not come with threads to easily add a filter nor do they generally list their diameter size. This makes finding the right size filter complicated and expensive.
  • It is hard to find surveillance filters and they are expensive. We paid over $150 for an IR bandpass filter and a mounting adaptor to make it fit with a regular surveillance lens.

1 report cite this report:

License Plate Capture 4K Test on Jan 11, 2016
License plate capture is one of the most popular surveillance applications....

Related Reports

Integrated IR Camera Shootout 2020 - Avigilon, Axis, Bosch, Dahua, Hanwha, Hikvision, Panasonic, Uniview, Vivotek on Jan 30, 2020
The best and worst cameras tested in this IPVM shootout showed major...
Cheap Camera Problems at Night on Feb 19, 2020
Cheap cameras generally have problems at night, despite the common perception...
IR Surveillance Camera Guide on Feb 06, 2020
Integrated infrared (IR) cameras are everywhere in 2020, but not all IR is...
The US Fight Over Facial Recognition Explained on Jul 08, 2020
The controversy around facial recognition has grown significantly in 2020,...
Coronavirus Hits Manufacturers, Standing Now, Worse To Come on Apr 06, 2020
Coronavirus is hitting security manufacturers, though overall modestly for...
Video Analytics 101 on Mar 16, 2020
This guide teaches the fundamentals of video surveillance...
Facial Recognition: Weak Sales, Anti Regulation, No Favorite, Says Security Integrators on Jul 07, 2020
While facial recognition has gained greater prominence, a new IPVM study of...
HLC Camera Shootout - Avigilon, Axis, Dahua, Hanwha, Hikvision, Panasonic, Uniview, Vivotek on Mar 02, 2020
Highlight Compensation (HLC) claims to reduce the effects of strong light...
Latest London Police Facial Recognition Suffers Serious Issues on Feb 24, 2020
On February 20, IPVM visited another live face rec deployment by London...
Video Surveillance History on May 06, 2020
The video surveillance market has changed significantly since 2000, going...
Integrators Hard Hit By Coronavirus on Mar 18, 2020
Integrators are already being hard hit by Coronavirus as brand new IPVM...
Fever Camera Sales From Integrators Surveyed on Jun 01, 2020
Fever cameras are the hottest trend in video surveillance currently but how...
Verkada: "IPVM Should Never Be Your Source of News" on Jul 02, 2020
Verkada was unhappy with IPVM's recent coverage declaring that reading IPVM...
Injes Tiny Temperature Terminal Tested on Jul 17, 2020
While temperature terminals have trended bigger, the Injes DFace801 is...
Converged vs Dedicated Networks For Surveillance Tutorial on Feb 12, 2020
Use the existing network or deploy a new one? This is a critical choice in...

Recent Reports

Google Invests in ADT, ADT Stock Soars on Aug 03, 2020
Google has announced a $450 million investment in the Florida-based security...
US Startup Fever Inspect Examined on Aug 03, 2020
Undoubtedly late to fever cameras, this US company, Fever Inspect, led by a...
Motorola Solutions Acquires Pelco on Aug 03, 2020
Motorola Solutions has acquired Pelco, pledging to bring blue back and make...
False: Verkada: "If You Want To Remote View Your Cameras You Need To Punch Holes In Your Firewall" on Jul 31, 2020
Verkada falsely declared to “3,000+ customers”, “300 school districts”, and...
US GSA Explains NDAA 889 Part B Blacklisting on Jul 31, 2020
With the 'Blacklist Clause' going into effect August 13 that bans the US...
Access Control Online Show July 2020 - On-Demand Recording of 45+ Manufacturers Presentations on Jul 30, 2020
The show featured 48 Access Control presentations, all now recorded and...
Face Detection Shootout - Dahua, Hanwha, Hikvision, Uniview, Vivotek on Jul 30, 2020
Face detection analytics are available from a number of manufactures...
Sunell is The First China Manufacturer to Market NDAA Compliance on Jul 30, 2020
Most China manufacturers are going to be impacted by the NDAA 'Blacklist...
Ink Labs Relabels China YCX Fever Camera And Steals Dahua's Marketing on Jul 30, 2020
A US company marketed a 'thermal temperature scanner' as its own, selling...
Genetec and Dahua-Backed Intelbras Split Examined on Jul 29, 2020
China is the cause of the breakup between Canada's and Brazil's largest video...
This YouTuber is Now Selling ThermoHealth Temperature Screening on Jul 29, 2020
An enterprising 20-year old is mass marketing medical devices on Facebook and...
Hikvision Returns To Growth Driven By Overseas Fever Cameras on Jul 29, 2020
While Hikvision's revenue fell in Q1 2020, it rebounded in Q2 attributed to...
Brazil's Biggest Domestic Surveillance Company Intelbras Profile on Jul 29, 2020
While Intelbras is not widely known outside of Latin America, Intelbras is a...
The Kiosk Market Pivots To Temperature Screening (Interviewed) on Jul 28, 2020
Video surveillance is not the only market that has pivoted to medical device...
Integrator Acquisitions 'A Good Market' During COVID-19, Says Greybeards on Jul 28, 2020
Industry broker Ron Davis of the "Greybeards" says that the integrator and...