By James Marcella, Director of Technical Services, Axis Communications
The physical security practitioner has a myriad of electronic devices at their disposal to detect, deter, and deny adversaries from harming people or property. Video surveillance is one such countermeasure used for deterrence and, more importantly, as an assessment tool to verify the events of a given scene and situation. The quality of that video could mean the difference between identifying a suspect or not.
Enter the megapixel camera.
Fueled by the insatiable appetite of consumers everywhere, digital camera, mobile phone, and television manufactures started a pixel count arms race that continues today with the latest buzz surrounding 4K, or ultra high definition television (UHDTV). The mantra, “More is better,” has made its way into the physical security industry with some equating the number of pixels to the overall quality of an image.
In video surveillance, however, more (pixels) isn’t necessarily better. The savvy security professional understands the value of surveillance video is based on its “usability” for the intended purpose of one or a combination of the following three use cases:
Each of these use cases has a desired set number of “pixels on target.” For instance, a general rule for identification is 80 pixels across the width of a person’s face. If certain factors line up, 80-pixels-per-face can easily be achieved with simple VGA resolution, let alone the eight megapixels defined in the UHDTV specification. The factors that influence pixels on target include camera sensor size, lens focal length, and distance to target. These factors are used to calculate a camera’s field of view (the horizontal and vertical area produced in the image referred to as FOV) which, can then be used to determine pixels on target.
Sounds complicated right? If you pick the right manufacturers, it shouldn’t be. Quality network cameras will have a built-in feature that enables you to determine pixel count from the computer by resizing a window around the intended item. This simple tool ensures that you have the appropriate pixels on target for the intended use case.
So when don’t megapixels matter?
If you wanted to identify individuals who entered the showroom of a car dealership you could set up a VGA network camera a certain distance from the door with a FOV that delivers 80 pixels or more across a person’s face. There is no need to buy a megapixel camera to achieve measurable results. That stated, with megapixel and HDTV technology, we can achieve a combination of detection, recognition, and identification from one camera. With more pixels produced by the IP camera, we could mount a lens that delivers a wider FOV while maintaining the appropriate pixels on target. Now the camera’s FOV includes the entrance way where we identify the person walking in, the car lot outside the dealership windows where we recognize people are looking at cars (See the article, “IP Cameras 101: WDR And Why It’s Important”) and the entrance to the lot where we can detect cars arriving.
With the prices of IP cameras dropping, many opt to purchase higher resolution cameras but continue to record or view at lower resolutions. This can be viewed as a good “future proofing” strategy enabling companies to increase resolution as the cost of storage drops over time or if their requirements change. Either way, the trend towards higher resolutions does not seem to be slowing down but it doesn’t mean we all have to jump on the bandwagon. Know your customer’s needs. Know your scene. And know the tools you need to calculate pixels on target.