Tuesday, December 20, 2011
The Grant County Sheriff’s Office (GCSO) serves a predominantly rural agricultural community of about 90,000 people spread over about 2,680 square miles in central Washington State. Unfortunately, we’ve recently seen some of the problems common to more urban departments, and we’ve faced some of the same staffing challenges. Patrol staff handles this huge geographic area in a continuous response mode, leaving little time for proactive efforts.
Specifically, Grant County has been experiencing a significant increase in the level of gang-related crimes, including drive-by shootings and homicides. Property crimes, including burglaries and thefts, finance drug abusers and gang members. Based on repeating offenses and substantiated gang involvement, we were able to pinpoint hotspots. If we could monitor these areas, we knew we’d be able to gather valuable information. High-risk events—such as fairs, concerts, graduations and gang-related funerals—needed a proactive approach that could also benefit from high-quality video surveillance.
A few years ago, we recognized the potential for high-quality photos and video to supplement our manpower and lead to a more proactive approach to policing our jurisdiction. Internet protocol (IP) megapixel video cameras were identified as a potential force multiplier when used at locations that we could readily identify. The idea was to temporarily place cameras at locations of criminal interest for short periods of time to gather specific information or document expected criminal action.
So what’s a megapixel camera? Basically, it’s a camera with a very high resolution (ability to “resolve” or capture an image with detail). This allows users to increase image size and pixel density to a level that makes images truly useful for identification purposes, rather than the blurry, unusable images we so often see from closed circuit television (CCTV) analog cameras at convenience stores or schools. High-end CCTV analog cameras may have a resolution of 640-x-480 video graphics array (VGA), or 307,200 pixels. A 1,080P camera (1,920-x-1,080 resolution) is 2.07 megapixels or 2,073,600 pixels. If the optics of the camera were designed to create the same pixel density as the VGA (640-x-480) camera, the image of the 1,080P camera would be 6.75 times larger than the VGA image. Put simply, you have a lot more information to work with when you use a megapixel camera.
For our law enforcement uses, we want much higher detail. This requires greater pixel density of the viewed area so the area to be viewed is reduced with the optical magnification of the lens selected. Lenses with more magnification will reduce the field of view. But it will also provide more pixels per foot of the image being viewed.
Arecont Vision, a manufacturer of megapixel cameras, says that a minimum of 30 pixels-per-foot is necessary for identification of a person. The company indicates that 45 pixels-per-foot is necessary to read U.S. license plates, and 100 pixels-per-foot is necessary to use facial recognition software. Others say you need more: 66 pixels-per-foot, for example, for license plates. Knowing what level of detail you want from your images and what pixel density that requires determines equipment selection and placement.
The above-mentioned pixel densities are for the horizontal camera resolution. The 1,080P camera (1,920-x-1,080 resolution) has 1,920 pixels on the horizontal plane. So, divide that number by the width in feet of the image within the field of view at the location to be observed to get the pixel density. You can do the math, or use charts and calculators provided by the camera manufactures. Remember: The magnification of a lens with the 1,080P camera must provide 45 pixels-per-foot to reliably read a license plate at your anticipated distance.
Complications occur when your magnification results in the field of view getting too narrow. Not only must the resolution be adequate, but so should the camera’s coverage area. If it’s too narrow, the solution requires placing the camera closer or changing to a higher resolution camera such as a three-, five- or ten-megapixel model. For law enforcements purposes, moving the camera closer often isn’t possible because the suspect would then become aware of the camera. We determined that a megapixel camera with optical zoom or interchangeable lenses would be necessary to fulfill our law enforcement needs.
In the early discussion of IP megapixel cameras, I consulted with our county IT director and staff. They were already using some lower-resolution Axis IP cameras for in-building purposes at county facilities, which meant they had an Axis video server and storage already established. They reported good results with the Axis products.
With funds remaining at the end of a gang grant, we purchased an Axis Q 1755 two-megapixel camera when they first came to market. This camera has 10X optical zoom and 12X digital zoom with auto focus. We liked the flexibility this combination offered. The optical zoom capability allowed us to easily address the pixel-density vs. image-size issues within the capability of the two-megapixel camera. The 10X optical zoom provided the equivalent of 5–50 mm lens capability. The digital zoom doesn’t help with pixel density, but it does provide additional zoom capacity (with the tradeoff of decreased detail). The camera supports H.264 compression, which we identified as necessary for any megapixel camera that we selected. The compression reduces the bandwidth required for video transmission and storage. This is an important consideration if you are transmitting the video.
The camera included a SDHC memory card slot that would allow for recording to a card plugged into the camera. Since the camera can be set to record only when there’s motion in the frame, the 32-GB capacity of an SDHC card provides a significant amount of on-camera storage. Used in this manner, the camera could be set so that once the card got full, the oldest video dropped off to allow new video to be recorded. A person must manually recover the memory card when video is to be viewed, but it does provide footage in areas where no network connectivity is available.
Along with the Axis Q1755 camera, we purchased a Sierra Wireless Raven X cellular modem. The limitation of this approach was the availability of adequate bandwidth on the cellular network. In areas with 3G sites, we could obtain 300–1000 kbps, which, when coupled with reduced frames per second and the H.264 compression, allowed us to stream the video to our server for recording and remote camera control.
The camera and the modem were placed in gray Hoffman electrical boxes that we modified with glass panes. Hoffman boxes are common and, from a distance, didn’t draw attention. They worked well for concealment on buildings, sheds, light poles or other existing structures.
We also purchased a pair of Ubiquiti 5-Ghz WIFI nano station radios that have a small form factor with integrated antennas. At $180.00 for the pair, the Ubiquiti data radios provided another way to backhaul the streaming video to points of fiber connectivity. The Ubiquiti Nano radios were recognized as a solution when our bandwidth needs exceeded what the cellular network could provide. In other words, if you need to transmit your video, these are a viable option.
The Axis Q1755 camera and associated equipment delivered high-quality images that met our goals. Following the successful pilot experiences of the first camera, we requested money and were awarded a gang enforcement grant that included funds for several cameras, a storage array and video-server software.
We were diligent in our review of what we wanted to purchase. We spoke with local organizations that were using high-end video. We attended security conferences, shifting through the sales and product information. We also attended vendor integrator training sessions on their cameras and software of choice.
With megapixel cameras on the market from multiple different vendors, we liked the idea of a software product that would support cameras from several different vendors. Most of the video software vendors provide potential customers the ability to set up their software and try it free for 30 days.
A lot of time was spent on the selection and evaluation of video surveillance software. We worked through several software trials and learned that the most expensive products were more expensive because they actually did what they said they would. Some of the smaller video software vendors claimed to support the new cameras coming to market, but didn’t have complete camera controls. Another company’s product wouldn’t deliver the recorded video in a timely manner if you wanted to see more than a minute’s worth.
Software that allows a person to immediately go to and view recorded video is a must. Many features that affect the ability of your officers to use the cameras and for your IT department to support the camera system are included in the software package. Electronic pan, tilt and zoom (PTZ) is an important feature with all megapixel cameras for zooming within recorded images. The ability to digitally zoom within recorded or live images is one of the clear advantages of high-resolution images.
Our IT department wanted a single software product they could use for all of the county’s video needs. In support of that concept, they wanted a product that would integrate with the county’s existing Active Directory structure. Active Directory is a Microsoft program that allows computer users to be assigned to multiple, different groups with specific permissions. The choice to have a software product that would support Active Directory drove us to the enterprise levels of software, adding substantially to our costs. However, this choice allowed adoption of a platform for much broader uses than ours alone. It also reduced the setup time compared to independently entering all of our individual’s permissions into a non-integrated platform. For agencies wanting to do only law enforcement video surveillance, software packages with limited numbers of cameras and reduced integration are significantly less expensive.
We spent a great deal of time eliminating unsatisfactory video software providers from our consideration. It’s important to determine your needs and the scope of your operation before you commit to an equipment purchase. If you don’t, you’re likely to either be disappointed due to limitations or spending more money than necessary.
Building on Success
After doing our research, we purchased Arecont Vision two-, five- and ten-megapixel cameras with 4–10 mm, 25 mm, 50 mm and 85 mm lenses. The Arecont cameras were significantly less expensive than Axis cameras of similar resolution, but they have some limitations. Arecont supports mostly fixed changeable lenses rather than optical zoom capability. The Arecont cameras aren’t autofocus—the focus must be manually set by the camera installer. On-camera SDHC memory storage doesn’t exist. Arecont cameras are limited to electronic pan-tilt: You can’t change the direction of the camera lens other than manually.
Although some of these differences are significant negatives, the lower cost and particularly the high-pixel density of the 10-megapixel camera provides real advantages for some circumstances. The 10-megapixel camera with an 85-mm lens (2.9-degree field of view) provides the ability to read license plates at 1,500 feet with a viewing area 75 feet wide. The manual adjustment of focus and direction are less of an issue if cameras are set in fixed locations and left undisturbed, such as in normal business or security deployments.
We also purchased two P5534 Axis PTZ cameras. These 720P cameras (1,280-x-720 resolution, just under one megapixel) include 18X optical zoom, 12X digital zoom with auto focus, SDHC memory card slot and power over ethernet. These cameras have proven themselves to be easy to set up and remotely control for exact coverage of areas of interest.
Image quality is excellent, although we often wished for larger image size when zoomed in. Axis is now marketing the Q6035, which is a similar PTZ 1,080P (1,920-x-1,080 resolution) camera with 20X optical zoom and 12X digital zoom. We haven’t had the opportunity to try this camera but believe that it addresses the one change that we had hoped for in the P5534 camera. The Q6035 camera also advertises the active gatekeeper feature of zooming in on detected motion and tracking that motion. This feature, if functional as advertised, will be of great benefit for law enforcement uses.
The selection of target locations for cameras must include consideration of adequate lighting. All of the described cameras have night mode, but require some lighting to be useful. Image quality degrades significantly after dark without adequate supplemental light. Infrared illumination is an option but the range of these units must be considered. Note: Cameras auto-adjust to bright light levels, but direct car headlights will wash out otherwise usable images. Camera placement and arranging for network connectivity takes time and patience. Make sure you look at resulting images in different lighting situations before committing to a fixed installation.
We’re happy with the quality of images available to us from our megapixel cameras. We recognize that we must simplify the deployment of cameras and backhaul, making it easier for field personnel to deploy. The PTZ auto focus cameras with higher megapixel resolutions and optical zoom capacity will go a long way toward that simplification of deployment. As LTE cellular or other broadband resources become available, this will also provide the needed bandwidth to better and more simply support this technology and approach to law enforcement.
How Many Pixels?!
necessary for ID Feature
45 U.S. license plate
100 Facial recognition software
Figures according to Arecont Vision.
Picking Software Vendors
Camera software companies may promise similar capabilities, but how they deliver makes all the difference. Take advantage of software trial periods. Run it through the paces and consider how police cameras may need to integrate with existing public safety cameras.
Jon Melvin has worked for the Grant County Sheriff’s Office as a deputy for 27 years, working a wide variety of positions within the agency. He holds Washington State career-level certification as a first-level supervisor. He currently works in boating enforcement and fire investigations, in addition to radio communications and technology work. Melvin works with the NIJ technical working group for communications. Additionally, he’s worked as an EMT, fire chief and search-and-rescue team member, and he’s trained for type-3 incident management team deployments.