Guest fathead Posted May 4, 2005 Posted May 4, 2005 Just my take on things: A common problem: monitoring a car park or driveway entrance some distance from the nearest position where you can fix the camera. You may be able to capture some video of someone breaking into a car but you have little chance of being able to identify the villain. From what I read this sort of thing is a big problem for the whole cctv industry. It appears that over 90% of the video submitted as evidence is inadmissible due to the poor quality of the images. The nub of the problem there is no way you can connect multiple cameras to a PC or DVR and capture high resolution video unless you use high compression and /or low frame rates. The PCI bus is just not up to the job. PCI Express doubles the throughput but that is still nothing near what is required for high res video. Looking at the spec on DVRs it becomes obvious that the frame rate has to be dropped to single figures to cope with data from many cameras. I am left wondering if this can really be called video. You might as well be taking a series of still pictures. Reviewing current alternatives it seems that the future will be IP cameras. High res ones are available but are frighteningly expensive. A large part of the cost of these cameras is no doubt due to the built-in computer and the compression hardware
secboy Posted May 4, 2005 Posted May 4, 2005 Just my take on things: Just one video camera plus the digital and you get it all. It is so blindingly obvious to a fathead like me that I can only conclude that I have missed the blindingly obvious. Someone please put me out of my misery and point out the obvious flaw that I have missed. 51237[/snapback] Sorry but I must suffer from that attention deficiency thingy! (I'm awake now) What are you trying to prove? Am I right in saying that you think that the industry need only manufacture one high res cam camera with a big enough varifocal range ,build in control and stick it in a housing? Now at first I wondered what you were going on about but now I think about it I think I see what you mean ! But that does not solve the problem of the equipment which then processes the image does it! or does it?. The resultant play back will only be as good as the dvr or multiplexer?. Ouch !!my head is hurting !!!! Have I lost the plot ? Because this does seem to make sense to me!!!! What does anyone else think cos if it does make sense I need to get an idea logged at the patent office quick!!!!!! Paul. (Sorry fathead I meant WE partner!!!!!!)
secboy Posted May 4, 2005 Posted May 4, 2005 Dave wots up with the clock its an hour out on the time i posted the reply?????Paul.
Guest Posted May 4, 2005 Posted May 4, 2005 Dave wots up with the clock its an hour out on the time i posted the reply?????Paul. 51246[/snapback] My Controls, (top right), set clock and time zone etc.... Wish I could read all of that post in one go, just off for another attempt!
secboy Posted May 4, 2005 Posted May 4, 2005 My Controls, (top right), set clock and time zone etc....Wish I could read all of that post in one go, just off for another attempt! 51248[/snapback] It was a bit long,but there may be some interest there,wonder wot some one in the tech support of a camera man'r might think???? Chears about the time--didn't think of the simplest!!!! Paul.
amateurandy Posted May 4, 2005 Posted May 4, 2005 I expect the military will have something clever that does all you want. At a price! The basic problem is that you want to collect high resolution data, but only when it's "interesting". You can't collect it all the time because the bandwidth/data storage requirements are too costly. And even if you could, you've still got to review/analyse it afterwards. So, you need 2 things: 1. A method of deciding what/where/when is "interesting" 2. A method of capturing high resolution image(s) of the interesting events Plus, probably, a continuous wide-angle recording to put it all in context. Traditionally you can achieve 1 by having a man watching monitors, and 2 by manually operating a zoom camera. But that process has lots of failings. So, you want software to do 1. CCTV guys, does this exist? It needs to be much more than a simple motion sensor. And as for 2, you'll probably need a panoramic view in high resolution with some sort of buffering for a few seconds, backed up by a zoom controlled by the clever software. Whatever, I don't think it's going to be cheap and easy. Or if it is it won't be effective.
Rich Posted May 4, 2005 Posted May 4, 2005 I don't really understand what you are trying to say fathead, but there's more than 1 factor to consider. Try looking at the professional range of DVR equipment. Cameras can have lenses fitted to enable them to be remotely zoomed in and out, the camera can be mounted on a PTZ unit so it can then be moved. From what I read this sort of thing is a big problem for the whole cctv industry. It appears that over 90% of the video submitted as evidence is inadmissible due to the poor quality of the images.What year is that information from, its more likely to be because the people designing and installing the system didn't have a clue what they were supposed to be achieving.The nub of the problem there is no way you can connect multiple cameras to a PC or DVR and capture high resolution video unless you use high compression and /or low frame rates. The PCI bus is just not up to the job. PCI Express doubles the throughput but that is still nothing near what is required for high res video. You can connect many cameras to a DVR, and record them all at a high resolution at high fps on a huge hard drive that can store 31 days of each camera. Also don't confuse the image you view on the screen to the image you are recording. When a digital image is transferred to an analogue PAL signal if the TVL of the camera and monitor match and you don't lose any of the information during recording, then the picture will look its sharpest. And with your idea about taking one huge image and then taking out the information from that image equivalent to a smaller resolution camera just seems more time consuming and more expensive, the video evidence when used in court has to meet a certain criteria, i.e. for identification the subject has to cover 120% of the recorded image size, not 120% of 10% of the original recorded image. Also don't forget the other most important part, the lens. There has been an idea similar to what you are thinking of, it was done at the Millennium dome, they took a high resolution camera with a wide angle lens and took spherical images of the inside of the dome, you could zoom in and out of the image to watch what was going on. What is your trade fathead out of curiosity?
Rich Posted May 4, 2005 Posted May 4, 2005 I expect the military will have something clever that does all you want. At a price! The basic problem is that you want to collect high resolution data, but only when it's "interesting". You can't collect it all the time because the bandwidth/data storage requirements are too costly. And even if you could, you've still got to review/analyse it afterwards. So, you need 2 things: 1. A method of deciding what/where/when is "interesting" 2. A method of capturing high resolution image(s) of the interesting events Plus, probably, a continuous wide-angle recording to put it all in context. Traditionally you can achieve 1 by having a man watching monitors, and 2 by manually operating a zoom camera. But that process has lots of failings. So, you want software to do 1. CCTV guys, does this exist? It needs to be much more than a simple motion sensor. And as for 2, you'll probably need a panoramic view in high resolution with some sort of buffering for a few seconds, backed up by a zoom controlled by the clever software. Whatever, I don't think it's going to be cheap and easy. Or if it is it won't be effective. 51251[/snapback] You can section off a screen into say 200 sections, then you can set the area that you want to use as a trigger, ie a doorway that shouldnt be used. Once the pixels in that selected area of the doorway have changed you can set it up to do a series of events. you can also set the desired size of the trigger objects. You can also track images moving on a screen. There really are alot more options for CCTV images nowdays that almost not much is out of the question any more. When using PTZ domes you can have them set up to do clever things also, like follow a running pattern and have a series of alarm triggers set up.
Brian c Posted May 4, 2005 Posted May 4, 2005 I'm with Rich and Secboy here. I'm not really sure what you're trying to achieve. If I understand you correctly, then the result you are after is already achievable. Like Rich said, motion detection can detect and track movement resulting from pixel changes. PTZ domes can have up to and over 22x optical zoom, together with the auto detection and tracking, this can provide images clear enough for identification. When a PTZ is displaying a 'zoomed out' image it may be 480TVL resolution, however when it's 'zoomed in' to a specific area, the resolution is still the same. If you take a 3Mpixel digital camera and take an image of a wide area, it will contain 3 million pixels, so if you concentrated on an area 10% of the image size, the viewed image would only be 300, 000 pixels. In other words, you would lose quality/resolution every time you 'zoomed in' compared with optical zoom. With a high-end DVR, the camera's can be set to record on their highest quality and frame rate (real-time in some), only when motion is detected. This helps keep with the space needed for the high quality images. Along the same lines though, anyone seen vista's new static PTZ!? It's a 360 degree fish eye lensed camera that uses digital processing to zoom in and distort the image to make it viewable. It is supposed to work like a PTZ but without moving parts. I'd be interested to see how it compares to an optical PTZ. If you don't know......ask.
Rich Posted May 4, 2005 Posted May 4, 2005 Interesting Bri, do you have a link to that vista camera/lens ? Im wondering if its the same thing they used at the dome.
Recommended Posts
Archived
This topic is now archived and is closed to further replies.