Video Surveillance, Access Control, IP Video Surveillance, Security News Network

Is analytics for IP cameras worse than traditional sensors or analytics for analogue cameras?

Is analytics for IP cameras worse than traditional sensors or analytics for analogue cameras?

What do you think is the influence of compression art efacts on the results of video analytics such as object tracing. Is it really true that due to artefacts you have an increased number of false alarms in your systems or is it simply enough to spend enough bandwidth and don't compress the images to much to have the same results? Is the situations going to get worse when using H.264 compression. What is you experience in real world projects?

Tags: analog, analytics, nvss, surveillance, systems, video

▶ Reply to This

Replies to This Discussion

Permalink Reply by neil on February 11, 2009 at 5:17am: An analog camera will deliver a better image (RAW) than an IP camera that make his own compression so you "lost data". Even more...depending on which compression, it could be very different.

I think that IP world and video analytics are not friend since the first one need to save bandwidth and second one need good quality (not resolution or FPS) video to analyze. It doesn't matter MPEG4 or H264 compression...if you compress, you loose quality on the image...and you loose "details".

Where it make sense is to embed video analytics on IP cameras or encoders...so analysis is done on the camera with the signal that camera receive...or if it's an encoder, from analog signal connected. In both cases, it's BEFORE compression...so, it's better quality.; ▶ Reply to This

Permalink Reply by Gilberto on February 16, 2009 at 8:56pm: I agree with you Neil, the cámera is the place to do the processing although you might want to keep some upgrade optional in the cámera and still keep some control in the mainframe as the algorithms will evolve overtime and you do not want to replace your cámeras which main operation (capturing and transmitting video) is fully functional.; ▶ Reply to This

Permalink Reply by wael badawy on February 18, 2009 at 2:32pm: I would say the following

Analytics is better perform on RAW data.
Analog Cameras will provide a better performance than IP Cameras if you apply the analytics at the NVR
Analytics will enable the deployment of HD (more than 1.3 MPixel) the problem of these resolution is you can not stream the images due to network bandwidth. Analytics will be the only enblers to process daya before compression i.e RAW format.

what do you think?; ▶ Reply to This

Permalink Reply by Gilberto on March 26, 2009 at 11:28pm: The only thing I do not agree with is that you can not stream HD. Bandwidth will become more available and HD streaming will be a reality for CCTV as it is for TV broadcasting.
Embedded modules will be critical for VCA on HD as a PC will not be able to process several cameras simultaneously. Embedded modules (with firmware upgrade) will be the key to support it. You might want to do VCA analysis at HD and only streaming D1 + some points of interest.; ▶ Reply to This

Permalink Reply by neil on March 27, 2009 at 6:08am: When you are streaming 32 channels+ over the internet it is a problem not on a lan setting.
neil; ▶ Reply to This