When it comes to selecting an IP camera the first question
out of some people’s mouths is how many megapixels does it have?
But before asking this question, you should ask yourself how
many megapixels do I need?
And before asking that the question make sure you know what
degree of resolution you need?
Do you want to see the flow of
foot traffic in your department store?
or do you want to see the facial features of
the woman pocketing that watch?
Resolution
Resolution is defined as the number of picture elements
comprising an image. When dealing with IP cameras, the pixel is the unit of
measure. Resolution may be expressed either horizontally x vertically
(640x840), or as a total number (1.4 megapixels). The finer the detail you want
to see, the greater the resolution (more pixels) you should seek. However it’s
not quite that simple. Megapixels do equal greater resolution but there are a
few more considerations, some plus, some minus.
(+) Megapixels add
greater resolution to the equation and you get a higher quality image.
(-) But now you need more bandwidth and have higher
storage requirements.
(+) Compression and frame rate adjustments can
solve this.
(-) But this could lower the quality of your
image.
Bearing all this in mind, the most practical thing to do is
decide how many megapixels you need without going overboard?
PPF
Camera developers have
come up with a minimum of 40 pixels per foot (ppf) as the standard for facial
recognition.PPF refers to the resolution of the final video frame and is based
on the size of the area being recorded. However Recognition is a broad term,
implying that this is a person is already known to the viewer. In most cases
recognition is based not only on facial characteristics but also on familiarity
with the subject’s body build and perhaps choice of clothing. If the person is
looking away from the camera or wearing sunglasses, or a hat is obstructing
their face, you may sort of, kind of think you recognize then but you can’t be
sure. The ideal resolution would allow you to identify them. Notice I made a
distinction between recognizing and identifying. For identification the standard
has been raised to 80 ppf (across the person’s face. So if you are watching a section of the store that
is 20 foot wide, say the jewelry counter,
you need a camera that gives a horizontal resolution of at least 1600
pixels wide (20 feet x 80 pixels).
40ppf 80ppf
On Second Thought
Suppose you want to cover the 20 foot jewelry counter plus
10 feet on either side, your total area of surveillance is now doubled to 40
feet. Picture a triangle here, by moving back you have now widened the base of
the triangle but the angle remains the same. You have widened the field of
view. And with a wider field of view comes the need for higher resolution to
maintain that optimum facial identification. Since you have doubled your area to 40 feet, using
that original camera with a megapixel resolution would now cut your final
resolution in half to 40ppf adequate for facial recognition but not
identification. So if you want to keep
that requisite 80 ppf, you could consider purchasing a camera with a higher
resolution (more pixels per foot, in this case, 40 feet x 80 pixels) of at
least 3200. But now you have those storage and bandwidth excesses we spoke of.
But there is another way.
Consider the Lens
If you remember from previous blog entries, the lens plays a major part in the quality of the image and
the field of view. It determines how far away you can see
something and it also defines how wide an area you can cover.
If we are monitoring an area closer to the camera we need a wider angle
lens with low magnification thus a smaller focal length (less mm). But if we
are monitoring something farther away, our angle of view becomes narrower and
we need a higher vocal length (more MM) which will give us greater
magnification. With a fixed angle lens this is an either/or proposition but in
the expanded area situation described above we need both. There is a way,
however, to widen the angle of view, maintain focus and still use the original
camera.
You could use a variable
lens. Variable lenses give you a choice of
fields of view. If the field of view is changed, focus can be maintained within
a range of focal lengths. Variable lenses are labeled according to zoom capability.
For example a lens that states having an 8x zoom refers to the ratio between its
longest and shortest focal length, and might give you a range of 6mm to 48 mm.
This would be an efficient move. You still have facial
recognition but since you have not increased the resolution, you haven’t
increased the bandwidth. Furthermore if in the future, a decision is made to
once again change the viewing area, this lens can adapt, giving you added
flexibility.
Carpenters used to have an old saying when it came to making
sure they weren’t wasting wood. Measure twice, cut once. The same could be
applied to choosing the right camera. Consider your measurements. Granted, choosing a camera without adequate
resolution won’t give you the surveillance you need, but at the other end of
the scale, purchasing a camera on the sole basis of megapixels can very well
give you more resolution than you need. This
means you’re not only wasting budget dollars but eating up bandwidth without
reaping any additional benefits.
If you need advice in choosing a camera or putting together
a system that will give you the most for your investment, call Kintronics at
800-431-1658 or go to our website and fill out a request form.
No comments:
Post a Comment