Camera Selection Guide 2013

Published Nov 14, 2013 05:00 AM

UPDATE: This has been replaced by Camera Application Selection Guide 2017

Entrances, hallways, rooms and parking lots are perhaps the 4 most common areas where surveillance is deployed. But what is the best type of camera for each? What resolution, form factors and FoV are best? Where do special features like WDR, integrated IR, super low light fit? Where is the best place to position the cameras?

In this report, we share test findings and recommendations showing tradeoffs of different camera options in each of these 4 fundamental spots - entrance, hallways, rooms and parking lots.

Below is an image of the FOVs used in our four different scenes.

Free Example

Inside, we have more than 20 comparisons. Here's a free example:

This comparison demonstrates the difference between mounting a single imager panoramic in the center of the room vs. a super wide angle camera at one end. 

Below is that same scene, just zoomed in on our subject.

More than 20 others inside.

Key Findings

Here are our key findings from this report:

Entrances/Lobbies

  • WDR cameras are essential in entrances with strong backlighting, a common occurrence. 
  • In smaller entrances (1-4 doors) standard definition cameras provide sufficient detail to identify a subject
  • Resolutions above 720p are unnecessary in most entrances and waste valuable bandwidth and storage space without adding to usable details delivered. These models also rarely include WDR.

Hallways

  • In hallways, both 4:3 and 16:9 resolutions waste much of the FOV on blank wall space, though 4:3 fits the scene moderately better due to its taller orientation.
  • Rotating cameras 90 degrees and using 9:16 corridor mode features greatly reduces the space wasted on unneeded areas and provides coverage of more of the hall.
  • Increasing FOV in corridor mode provides video of more area beneath the camera which may help to show direction of subjects. However, this increase results in a loss of PPF in the hallway itself, detrimental to recognition capability.

Interior Rooms

  • Placing a single super wide angle camera (134° FOV) in one corner of the room provides approximately the same details as a similar resolution panoramic camera. However, in our tests, wide angle cameras provided better light handling capability, resulting in less overexposure on our subject and chart, while still covering the entire room, making them preferable to panoramic cameras.
  • For average size conference rooms (20'x30' and below), resolutions higher than 1080p offer few additional details. 720p resolution provided slightly less than identification quality video, though a known subject may still be recognized.
  • Integrated IR cameras provided higher quality low light images than non-IR models, providing details of our subject and chart below 1 lux where others produced no usable images.
  • Placement at front or rear of room depended purely on what activity was most desired to be captured: entry/exit or general activity.

Parking Lots

  • 5 and 10 megapixel cameras provided limited benefits in outdoor scenes due to reduced light handling capabilities. Reflected light caused areas of the image to be washed out, including subjects, vehicles, and our test chart.
  • FOV depends on what details are desired. Wide angle views provided overview of activities in the parking lot, but a reduction of recognition of subjects and license plates. Reducing the FOV by about half provided better recognition capabilities, but a full overview of activity (cars and people coming and going) was lost, possibly requiring additional cameras in the same scene.
  • Since most parking lots have some form of outdoor lighting for personnel safety, super low light cameras provide details similar to integrated IR cameras. In unlit or very dimly lit parking lots, integrated IR models provide better performance.

Recommendations

Based on these observations we have the following key recommendations for each of these areas:

Entrances 

In typical entrances, we recommend SD or 720p cameras with good WDR performance (if strong backlighting exists). Cameras beyond these resolutions are not necessary unless the entranceway is very large, resulting in wasted storage and bandwidth. Cameras without WDR will lose detail due to backlighting, reducing their ability to identify subjects.

Hallways

In hallways, more resolution is not always better. While higher resolution provides more pixels per foot, these cameras also generally handle light worse than 720p or 1080p models, reducing usable image quality. If light is bright and even, and storage space is not a major concern, 5MP may be a fair tradeoff.

Where possible, we recommend corridor mode be used, shifting the aspect ratio from 16:9 to 9:16. This eliminates much of the dead space seen when using standard 4:3 or 16:9 cameras. However, this capability may be limited by camera or VMS choice.

Interior Rooms

In typical interior spaces such as conference rooms, we recommend at least 720p resolution be used. Lower resolutions fail to provide usable video. By contrast, high resolutions such as 5 and 10 megapixel may simply waste space unless fine details of subjects and activities are absolutely necessary.

Super wide angle cameras are recommended if the entire room must absolutely be covered, instead of simply one end or the other, due to their improved light handling capabilities compared to panoramic models centrally mounted.

Integrated IR cameras are recommended if activity is expected when lights are low. Non-IR cameras provide little in the way of detection or recognition of subjects.

Parking Lots

1080p cameras provided the best tradeoff between light handling and resolution in this scene. High megapixel cameras may provide more pixels per foot in parking lot scenes, but users should be aware of possible overexposure issues common in these models. Low light performance of cameras above 1080p is also poor.

We recommend wider FOVs be used in parking lots, as details at the ranges commonly necessary in these areas are low. Narrowing the FOV by nearly 50% does little to increase details on target at longer ranges, but reduces the camera's ability to capture an overview of activity throughout the scene.

Since most parking lots are lit, the decision of whether to use integrated IR or super low light cameras comes down to two factors:

  • Color: If color is a key need in the scene, in order to identify vehicles or subjects, low light cameras should be used.
  • Cost: If color is not needed, integrated IR cameras may generally be purchased for a lower price, while still providing details on par with low light models.

Entrances

First, we tested one of the most common locations for cameras in any building, an entrance lobby. For entrances, like the one pictured below, we tested the following:

  • WDR vs. non WDR
  • SD vs. HD vs. Multi-Megapixel

WDR vs. Non-WDR Cameras

Wide dynamic capability is essential in entrance scenes to compensate for strong backlighting caused by the sun outside the building. In the comparison below, with WDR on, the subject, chart, and area outside the building are all clearly visible. With WDR turned off, the subject's face is partially obscured, and outdoor areas are completely washed out.

SD vs. 720p vs. 1080p

In a field of view this narrow (~14'), even standard definition (704x432 in this case) is sufficient to provide recognition quality images of our subject and chart. Higher resolutions do not provide a practical increase in usable details.

The same is true outside the door, as well. Note that in this case, the 5 and 10 megapixel cameras' lower performance light handling negatively impact image quality, with our subject and chart both more washed out.

Hallways

For hallways, like the one pictured below, we tested the following:

  • HD vs. 5MP vs. 10MP
  • 4:3 vs 16:9 vs. Corridor Mode
  • Corridor Mode Standard vs. Wide vs. Super Wide FOV

HD vs. 5MP vs. 10MP

With the subject at the near end of the hallway, not surprisingly, all resolutions are able to capture quality video. Even 1080p at this range is overkill.

At the mid point of the hallway, ~30' from the camera position, details begin to drop, though the 720p camera still provides quality suitable to recognize our subject. Note that the 1080p camera in this case the 1080p and 10MP cameras started to suffer from greater digital noise and artifacting than 720p and 5MP models, reducing image quality.

At the far end of the hallway, both 720p and 1080p provide PPF too low to identify our subject or make out our chart. Digital noise issues are more pronounced at this range.

4:3 vs. 16:9 vs. Corridor Mode

Since hallways are much longer than they are wide, both 4:3 and 16:9 aspect ratios result in substantial amounts of space in the FOV wasted on useless areas such as blank walls (highlighted by the red boxes below). Corridor mode shifts this aspect ratio to 9:16, providing more coverage of useful areas, as can be seen highlighted in green below. 

Additionally, since it simply rotates the camera 90 degrees, corridor mode maintains the same PPF as standard aspect ratios, as can be seen as a comparison of the two below:


Note that corridor mode is not available in many cameras or VMSs, limiting its support. Users should see our test of Axis Corridor Format for more information on the tradeoffs of using 9:16 aspect ratio. 

Effects of FOV on Corridor Mode

To see the impact field of view has on corridor mode cameras, we tested it at four different fields of view, ranging from relatively narrow to super wide (134°). In the leftmost FOV below, the camera maintained a field of view similar to the 16:9 cameras above, about 65-70°, but was rotated into corridor mode. FOV was progressively widened, to ~90°, 118°, and 134° in images to the right.

Note that the area beneath the camera becomes more and more visible as the FOV is increased, which may be useful in determining direction subjects traveled as they move through the FOV. However, the very wide (118/134°) fields of view offer little more practical value, as direction can be determined even on the narrowest field of view.

Additionally, details drop quickly as the FOV is widened, as can be seen in the image below. In this case, ability to identify our subject midway down the hall becomes lost at 90° and beyond.

Rooms

For rooms, such as the conference room below, we tested the following:

  • Camera location: front vs. rear
  • Corner mounted wide angle vs center ceiling mounted panoramic
  • SD vs 720p vs 1080p
  • Minidome vs Box
  • Integrated IR vs color only

Front vs. Rear Mounting

The first question in locating cameras in interior rooms such as these is the camera's primary purpose.

  • Mounted in the rear, the camera can be used to view general activity in the room, but a view of entries/exits through the door may be lost.
  • Mounted in the front of the room, the camera may more easily capture subjects entering and exiting, but activity in the front of the room may be missed.

These two shots show the difference between front and rear mounting:

These differences may be mitigated by using super wide angle or panoramic cameras, though at a loss of PPF throughout the FOV.

HD vs. Multi-Megapixel

In this typical conference room (~20' x 30'), we found 1080p to be the approximate "sweet spot" in resolutions. 720p was not quite sufficient for identifying video of our subject's face. Increasing resolution to 5 or 10 megapixel added few practical details not seen in the 1080p camera, but reduced low light performance substantially, as seen below.

In this scene, our 720p and 1080p cameras offered by far the best performance. The 720p model was modestly more able to discern letters on the test chart compared to the 1080p camera, due to decreased digital noise. Neither the 5MP nor 10MP cameras provided usable images of our subject or chart.

Super Wide Angle vs. Panoramic

This comparison demonstrates the difference between mounting a single imager panoramic in the center of the room vs. a super wide angle camera at one end. 

Below is that same scene, just zoomed in on our subject.

Integrated IR vs. Color Only

Finally, we tested two 720p cameras, one color only, and one with built-in IR, to see what benefits, if any, IR illumination offers in this scene. This image compares the field of view of both. In the non-IR camera, there is some difficulty detecting our subject, though the chart may be easily seen. The IR camera makes detection much easier.

Zooming in on our subject, we can see that neither camera provides recognition quality video at this distance. 

Parking Lots

For parking lots, like the one pictured below, we tested the following:

  • HD vs 5MP vs 10MP
  • Super wide vs moderately wide
  • Integrated IR vs Super low light

HD vs. MMP

At this distance, across an approximately 80' field of view, 720p resolution does not provide images sufficient for recognition of our subject. Additionally, the chart is barely legible past the first line. 1080p and above provide better recognition of both subject and chart. However, the 5 and 10 megapixel cameras are washed out starting at line four, where the 1080p camera is not.

Wide FOV vs. Mid

In this test, we narrowed the FOV from about 80', which captured the entire width of our parking lot, including the driveway to the right of the cars, to about 45', increasing PPF across about 5 parking spots. This reduction removes the ability to observe vehicles entering and exiting, which is captured in the wider FOV.

Additionally, reducing FOV does not result in much practical gain in details, with our subject and chart only modestly more discernable.

Integrated IR vs. Super Low Light

Finally, we tested an integrated IR camera against two low light optimized models to see tradeoffs in performance at night. Note that in this scene, the parking lot was illuminated to about 2 lux on target, typical of many parking lots. This image shows the difference in performance across the FOV, with the IR camera's illumination pattern evenly spread throughout.

At ~15', performance between low light cameras and the IR bullet is similar, with facial and chart details easily discernable. The 1080p camera in this case is somewhat washed out due to the outdoor light, resulting in some loss of detail.

At long distance, the cameras once again offer similar performance, though none are able to deliver much detail due to reduction in PPF and digital noise caused by increased gain.

Comments are shown for subscribers only. Login or Join