CCTV: The truth about H.264 and image degredation

The third part of the case for H.264 by Mark Harraway, UK Country Manager

Q. As an H.264 adherent could you please fight its corner in terms of image degradation.

Again, this is all about how far you go with the compression or profile used.  Many vendors offer functionality within their products to support additional compression or bandwidth limitation. In this way they improve the tailoring of their product to specific projects.  For example, if you are using remote cameras over an ADSL line with a fixed maximum available bandwidth, then it could be acceptable to either drop frame rates or decrease image quality in order not to exceed the available bandwidth and avoid pixilation.

The H.264 standard sets the criteria for multi-picture inter-picture prediction which is an improvement over the previous standards of just examining the differences between one or two frames by sampling many more frames. If it is unlikely that certain images areas will change (fixed structures in the background for example), then bandwidth is not wasted by repeated transmission of redundant data that is in the frame already. This does mean that bandwidth can be very dynamic in locations with frequent motion or sudden scene changes as the algorithm has to deal with such variation. But theoretically, depending on the profile used, there should be no difference in image quality. And if the hardware encoder then makes a choice to drop frames or over-compress the data to meet a user defined requirement, the H.264 standard should not take the blame.

Part 4 continues tomorrow…..


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: