I have been using the LRC Web site, for about a year now, and find it extremely helpful. The Links are a very convenient way to connect with manufacturers and I keep a library of your answers to questions. I have found that very useful.
I do have a question that I've been trying to research. In fact, that was the reason that I visited the LRC Web site today.
I've been tasked with comparing Digital Video Recorders on paper. I've obtained spec sheets for about 10 company's and even have a spread sheet prepared by Mr. Vlado Damjanovski which compares a multitude of specs for seven different recorders. I am having a problem with making a comparison with the compression rates for the different types of compression technology (JPEG, Wavelet, MPEG, H.263, and proprietary hybrid MPEG/Wavelet. I want to make a comparison that would be fair, proportional and in layman's terms if possible.
Most of the people that I deal with are used to a standard of "NTSC 30 fps" as real time CCTV video. I can explain that to everyone as, "It takes two fields (horizontal and vertical) to make one frame, and 30 fps/sec to make composite analogue NTSC CCTV signal." Now when it comes to the compression rates I get confused, and so do they. I've found several different terms used, such as frames, fields, images, pictures/per second and even "ms in 40 ms steps to 1000ms" which makes no sense to me. Is there anything available that will show some type of comparative spread between the various technologies?
Thank you for sending me the booklet on CCTV basics.
Sign me, Compressed and Confused
Dear C&C,
First things first, I apologize for taking so long to respond to your
excellent confusion. I have been chin deep in alligators and unable to see the
swamp. At any rate, you are actually approaching three separate concerns with
digital recording. 1) Compression engine 2) Resolution of the final product 3)
Frames per second. Each is a concern, but they should be approached separately.
So let's attack.
Compression: This is a rough one.
Currently, we have no standards over which compression standard should be used
for video. The ones that you mentioned were all originally designed for internet
transmission of graphics. Computer Graphics are a whole different ball-game from
Video. As you stated, video is made up of two half pictures at 60 fields / 30
frames per second, while graphics are a single grid of pixels at fixed ratios
4/3 (i.e.; 640 X 480, 800 X 600, 1024 X 768, 1280 X 960). With Video, we count by
lines of resolution ... how many lines paint the picture on the screen. The more
lines the better the detail of the image. With Graphics, the more pixels crammed
onto a screen, the better the resolution or detail of the image. To compress
video, we must first turn it into a graphic image. This by itself can cost as
much as 25% of the initial video image's resolution. Then we take whatever we
end up with and compress it into a smaller file to take up less space. How we
compress or what engine we use will determine the quality of the product when we
go to use it later.
Imagine a picture that is 8" by 6" large. It is a excellent picture
with a high resolution (good detail). This picture is made up of dots that are
arranged in rows and columns. We have 800 columns and 600 rows of dots. Now, if
we need to blow the picture up at this point, we can make it twice as large and
still have good detail. However, let's compress it first. We will make it
1" by 3/4" large. The only way to do this is to cut down the number of
columns and rows of dots. Now we have 100 columns and 75 rows of dots. OK, we
store the image in a much smaller area. We can even view it as a 1" by
3/4" image and it appears to be very sharp. However, now we want to make it
2 times bigger than 8 X 6 again... about the only thing we see is large dots
spread across our screen or paper.
Based upon the above example, the type of compression and the amount of
compression that you use is very important. Which one is better ... who knows,
everyone claims to be using the best. I do know that any compression of 20% or
more in JPEG will cause too much detail loss to be of use. The biggest key to
compression engines is which one will be around in two or three years and which
ones will fall by the way side.
Resolution of the image: Based upon the
above, you can see what compression can do to resolution. Consequently, the
resolution of the final image becomes important as based upon several different
items. 1) The size of the initial image. The larger the object of concern (in
the image) the less resolution will be needed to reproduce it. The smaller the
object of concern, the more resolution will be needed. Consequently, if you are
looking at a person at the end of a football field and that person appears to be
1" tall in a nine inch screen, you better have (and keep) as much
resolution as possible if you intend to be able to identify the person at a
later date. Bottom line, either do a better job with the design of the
camera/lens or do not use a compression storage method. On the other hand, if
you are using an image of a person that was only a few feet from the camera and
their face and/or body takes up the majority of the screen, the resolution of
the beginning and/or the final product is not as much of a concern.
The second argument on resolution will be dependant upon the purpose or need
of the final image. If you are using the stored image for the purpose on
counting cars that pass through a gate over a specific period of time, the
compression rate will be of no consequence. All you are interested in is how
many cars pass through the gate ... count the shadows. On the other hand, if you
intend to or could end up in a situation where someone's welfare, safety, and/or
conviction relies upon the detail of an image, you must be very care how much
compression is used to store the image.
Images per second: We have determined
that a minimum of 25 full frames of video per second will constitute real time
or constant motion imagery. In the US, and other areas where we live with 60
cycle power, we work with thirty pictures per second. For years, we have stored
these images onto magnetic video tape ... all thirty per second. However,
because of the amount of room that these images take up on a disk or magnetic
tape or chip or whatever, we are now being told that we don't need thirty
pictures per second ... we only need two or five or what ever. The facts are
that each job is different and each job can have several different aspects to
it. If I am trying to count the number of people passing through a gate, two
pictures per second may not be enough ... I might miss a person. On the other
hand, do I need thirty pictures, per second, of an empty room? Of course not.
Since the introduction of digital storage systems, we have been forced to design
our systems more conservatively, more carefully, more in advance. Since we are
limited on space for storage, we must be conservative, but realistic on how we
use our space. Think the application through and apply each camera to it's own
circumstances. Everything can be tested on paper prior to being put in place.
At any rate C &C, I hope that this helps you out a bit. I also hope that
it didn't get to you too late to be of any use. If there is anything else that I
or my staff can do to assist you in your CCTV needs, feel free to holler at any
time. We are in your service.
Sincerely, Charlie R. Pierce President LRC Electronics Company LTC Training
Center Niche Products Catalog Electronic Service Group
