Our Man in the Field: The Process of the IP Solution, Part X

Charlie takes an in-depth look at the common IP-video compression schemes


In my last column I opened the door to compression engines but didn't really go into much detail about the differences in the ones that we hear about every day. So, let's dive in and see what we can find.

Repeating myself somewhat, we have several different engines that are available to us in the CCTV market. Each one of these compression formats is a standard in the computer world. However, as is true with every good recipe, you’re not really cooking until you put your own variation to it. Therefore, if a manufacturer says that they’re using MPEG-4, because it is the only really good compression scheme out there, what they mean is that they have taken the MPEG-4 standard and modified it to fit their interpretations of your needs. Are their interpretations accurate? Who knows?

Let's start with JPEG. This is an acronym that stands for: Joint Photographic Experts Group. Pretty cool, huh? This compression engine records every frame of video or every picture. It compresses by discarding the fine details of the image. To do this, it takes combinations of pixels and combines them. If you have a block of four squares (pixels) and each is a different color, the JPEG engine will combine the four pixels into a single one that is a rough average of the four colors combined. That is to say, if you can see the hairs on a man’s head, the resulting compression in JPEG would take away the individual hairs and give you a guy with dark hair ... sort of. The amount of compression in JPEG is determined in percentages. You can compress an image from 1 percent to 99 percent. Each percentage of compression relates to the percentage of fine detail that will be glazed over or combined. If you compress an image by 50 percent, it will literally appear to be 50 percent smaller on the screen. Please understand that this means 50 percent of the horizontal and the vertical, so the image is greatly reduced. Looking at the smaller image, it appears that nothing has been lost.

But then the time has come to cut brass tacks and enlarge the image for closer scrutiny. What you will notice, more and more with every enlargement is what is referred to as the "Jug head" affect. That is to say the edges of various objects in your image will become blocks or squares. There is no room for a curve or slow radius or circle in JPEG. Motion JPEG or M-JPEG is a variation of the JPEG standard. So are all other formats that have the JPEG mantra in them.

Wavelet is a simple adaptation of the somewhat shaky JPEG 2000 standard and it has literally an infinite possibility of variations. This is because anyone can write a wavelet algorithm and everyone has. The key difference between the wavelet compression and its cousin JPEG is that wavelet does not tile. That is to say that it does not combine blocks of pixels into a single pixel. It looks at the frequency of the whole frame or image. High frequencies are noise and low frequencies are black. It applies combinations of these frequencies from the bottom of the image up. The net result is that as you magnify an image that has been compressed with a wavelet format, the image appears to go out of focus as opposed to becoming blocks.

The bottom line is that wavelet engines are capable of doing massive compression, but you still loose the fine details in the image. As for what wavelet stands for in the world of acronyms, I'm not really sure. I think it was thought up by some surfer dude while shooting a tube. [Editor’s note: It’s actually referring to a mathematical principal of the decay of oscillating waveform, though the surfer reference is easier to remember – kind of like a crashing wave.]

MPEG which stands for Motion Picture Experts Group is one of the more pronounced forms of video compression is this day and age. We have the MPEG 1, MPEG 2, MPEG 3, MPEG 4, MPEG 7, and MPEG 21. For the most part, video uses MPEG 1, 2 and 4.

This content continues onto the next page...