Is there an advantage of using a high resolution (570 lines) camera with a time lapse recorder that has a resolution of 300 lines?. In other words, would the recorded image from a 570 line camera be higher quality than a 380 line camera? How does the video recorder interpolate or "compress?" the higher res image into a lower res image? Is it similar to audio where a higher quality source makes a better recording than a lower quality source, even though the limitation of the recording device may be less than the higher quality source?
Thanks for your help. Just sign me "Losing lines in MO"
Dear lines
After all the hub-bub that is made about resolution, what is this phenomenon
that we keep telling you is so important to the video image? More importantly,
what is the real affect of high or low resolution images in our security
applications? I mean, we all know, or have heard, that you should use a higher
resolution camera outside, or for long shots, but why? In the end, does it
really matter what style of camera we use? Actually it does, but not necessarily
for the reasons that you have been thinking or lead to believe.
First let us describe and define resolution. There are several ways to
measure the resolution of an image. You can use the actual number of horizontal
lines used to manufacture the image (horizontal resolution), pixel resolution,
or vertical resolution. To understand any of the different resolutions, we must
first go back to the basics. First, how is the video image recreated on the
monitor screen? Don't run, this sounds highly technical, but we can break it
down to simple terms. To give an image the illusion of motion, we flash a series
of fixed pictures on the screen at a rate that is faster than the human eye can
detect. Each fixed scene is incremented slightly farther ahead in the action
than the one before it. When we put the image on the screen, we don't just flash
up a picture, we paint the image one line at a time. This line of video
information is painted on the screen from the left side of the monitor to the
right and is referred to as a horizontal scan line. We first go down the entire
screen painting only the odd lines, then we repeat the process filling in the
even lines of information. This process is referred to as two to one (2/1)
interface. In the end, we create a total of sixty (60) half images (fields) or
thirty (30) frames per second (US) or fifty (50) fields / twenty five (25)
frames in countries using 50 Hz power. The total sum of these lines create what
we refer to as the vertical resolution of the image. I realize that this seems
somewhat confusing, but vertical line resolution refers to the total number of
horizontal lines. This is because we are counting the individual lines of
information from top to bottom or vertically. Leave it to an engineer to make
life confusing.
In the US, Canada and many other parts of the world, we work under NTSC
(National Television Standards Committee) standards which fixes the video image
at sixty (60) fields per second and 525 vertical line resolution (525/60). In
Western Europe, Australia, parts of Africa, and the Middle East, they work with
PAL (Phase Alternating Line) format which fixes the video image at fifty (50)
fields per second and 625 vertical line resolution (50/625).
But how do we paint an image on the screen? What magic could possibly exist
that gives us the ability to make an image? It's really very simple. If you have
ever pushed your face close enough to a monitor or television screen, you have
seen rows of independent bright spots. These little spots are referred to as
pixel points. In the end, a video or television image is nothing more than a
collection of variously shaded pixel points. The total sum of these pixel points
on the screen is referred to as the Pixel resolution. The total number of
vertical columns of the pixel points (counted from side to side) is referred to
as the Horizontal resolution. This is because we are counting from left to right
or horizontally. Again, just a little confusing, but par for the course. The
horizontal resolution is determined first by the camera, second by the
transmission method, thirdly by the weakest link in your video system.
The combination of vertical and horizontal lines create an overall format
resolution that will determine the quality of the image. Since the vertical
resolution is fixed at either 525 lines (NTSC) or 625 lines (PAL), we depend
upon the horizontal resolution to determine or define our image quality. Imagine
the entire video image on the screen as nothing more than a drawing made on a
grid paper. The smaller the squares, the more defined the lines in your drawing
can be. Consequently, the more lines used to make the grid, the smaller the
squares.
Now, to the heart of the matter. The first lesson to remember is that the
overall quality or resolution of your image will only be as good as the weakest
link in your video system. Consequently, there is a lot of money being spent on
high resolution equipment, but the results are coming out equivalent to lower
resolution images. For example; the average nine inch (9") black and white
video monitor will reproduce, up to, six hundred lines of horizontal resolution.
As a quick note, the average television only produces three hundred and
twenty five lines of horizontal resolution, which is why a video monitor almost
always looks better than a television. Now, let's say that we have a camera
that produces an image with eight hundred lines of horizontal resolution. Will
the monitor make the eight hundred line image or will it drop off two hundred
lines of resolution somewhere and reproduce an image of six hundred line
quality? Answer, drop off two hundred lines of resolution and reproduce an image
of six hundred line quality. In retrospect, what if we had a camera that
manufactured an image of three hundred lines of horizontal resolution. Would the
monitor reproduce at six hundred lines? Answer, no. In the end, the monitor is
an idiot box with no ability to recreate what wasn't there in the first place.
The monitor however, is seldom the problem. Consider that the average video
recorder, used in today's surveillance video systems, only has a reproduction
capability of an average of three hundred and twenty five (325) to four hundred
and eighty (480) lines of horizontal resolution in the playback mode. "But,
my monitor has six hundred (600) line capability and I bought a special camera,
for a small fortune, that has eight hundred (800) line horizontal
resolution!" Oh well, now you know why your video playback is never as good
as the image on the screen. Figures, doesn't it. You thought that you had a
problem with your video recorder and have just come to find out that you really
do. You spent too much money on a camera that the VCR cannot emulate.
You have however, also just learned another key reason for using
professional, industrial video recorders instead of the two or three hundred
dollar consumer units that you buy at the store. It's really simple if you know
just one more thing and that is that your television only has three hundred
twenty five (325) line horizontal resolution. Therefore consumer recorders are
not going to be built at such a level as to produce more resolution than the
television can.
One place where we need to use higher resolution cameras is outside. This is
because the majority of our outside cameras are using longer than usual lenses
(more telephoto) and are usually on longer cable runs. Since we can loose
resolution or quality of image by looking through telephoto lenses, and we can
loose resolution on long cable runs, we prefer to stick with higher resolution
cameras up front. Think about it. If the image is going to loose resolution,
before it hits the CCD (Charged Coupled Devise), or because of the cable run,
wouldn't it make sense to have as much of this stuff as possible? Absolutely!
But this isn't the sole or even the most important reason for using the high
resolution cameras.
OK it's time to get to the meat of the situation. Assume that our position is
one of security and that we need to identify various objects or individuals on a
video screen. Also assume that we will be doing this from two perspectives. The
first perspective is while looking at the screen on the monitor with a direct
feed from the camera and the second is during the playback of the image off of
the video recorder. For the first example, I will use a camera with a horizontal
resolution of six hundred lines (a really, really good one). I am looking at a
six foot man and he fills my screen from top to bottom, head to feet. So we can
say that our image resolution is about five hundred and seventy (570) lines
(assuming a 5% lose due to cable and lens).
This is good, however, what is the amount of resolution that we can use for
the identification of the person from a facial perspective? Five hundred and
seventy lines? No, in fact the average human head (face) takes up only one fifth
(1/5) of the overall body. Therefore the true resolution of the mans face is
only one fifth of five hundred and seventy lines or roughly, one hundred and
fifteen (115) lines. This is still a huge amount of video information and so we
will be able to make a solid identification of the individual (provided that
lighting and angle of view are adequate). On the backside however, let's say
that we are now looking at the play back of our time-lapse video recorder. We
have the same man, same scene, same amount of scale and size. The difference
here however is that our recorder only has three hundred and twenty five (325)
lines of resolution at play back. Now what is the true resolution of the man's
face? One fifth (1/5) of three hundred and twenty five or about sixty five (65)
line resolution. Still very good and easy to identify, so we're OK.
For the second look, let's drop the man's overall appearance in the scene
from taking up the whole screen into a little more realistic attitude. Let's
make the man one inch (1") tall on an eleven inch screen (11")(as
measured diagonally). This would mean that his whole body takes up about one
sixth (1/6) of the overall screen. What is the overall resolution of the man
this time (assuming the same scenario as described above). For his overall body,
we would take one sixth of five hundred and seventy (570) lines of resolution or
we would say that his body had ninety five (95) line resolution. His face would
be one fifth (1/5) of that or nineteen (19) lines of resolution. Getting low,
but we should still be able to identify this individual. Now let's go to the
playback at three hundred and twenty five line resolution and take another look.
Our first calculation says that the man's body uses one sixth of the overall
scene for a total of fifty four (54) lines of resolution. His face is one fifth
(1/5) of that or roughly ten (10) lines of resolution. Probably still enough for
identification purposes, but certainly pushing the envelope. On the other hand,
what if we needed to identify something about this man in specific to make our
case.... say his shoes. His feet would represent about six inches of overall
identifiable surface area in real life. If we reduce this to screen size, we
could say that the top surface area of his shoes (assuming we have an elevated
angle of view looking down at the man) would represent about one third (1/3) of
the same amount of area that his face did. Therefore, we would have roughly
three and one half (3.5) lines of resolution, on playback to use for shoe
identification. It would not be enough to make a case.
At this point, you may be saying that I am going too far and that a man's
shoes is not a good basis for an example. I would wager that you make this
statement based upon the lack of believe that such is evidence. However, think
back to O.J. Simpson... were not his shoes of some question because of the
bloody trail? Several weeks ago, I was involved with an FBI case where they were
trying to prove guilt of a man based upon a video tape playback and the nature
of his shoes in particular. This man had entered into a large retail store, was
dressed in a black coat, black pants, black pullover stocking cap (not that
unlikely in Iowa in the winter) and black shoes with some sort of white markings
on the top. Turned out to be white doves on top of his shoes and they were the
ones he was arrested in. The problem came in the playback of the video that
captured his image during the robbery. When I first saw the image, it was played
back via a consumer recorder on a color television. His shoes were defined by
about two lines of resolution and the doves appeared to be something related to
what a dove might leave on your shoulder. Played back on a proper industrial
recorder and with a black and white monitor however, his shoes in fact had a
full six lines of resolution and the doves stood out like an accusation.
In the end, resolution of an image is much more than just how many lines we
use to paint the screen. It relates to the overall screen, relative size of the
object of identification, and the type of equipment that the image is recorded
or even played back on. This is what resolution really means to you in the
field. Therefore when you design your systems, keep in mind that the size of an
object in perspective to the overall size of the scene is also relative to how
many lines of horizontal resolution will be available to describe the object.
The smaller the object of identification, the higher the camera resolution will
need to be. The larger, less defined the object of view, the lower the overall
resolution can be. Keep in mind however, that it is the weakest link of your
system (usually the recorder playback) that will be the most important thing in
the end.
