CCTV Test: 360 Degree Cameras (Part 2)
Whilst many think that solid state 360 degree cameras are a new innovation, they’re not. The first video surveillance camera using a fish-eye lens and dewarping software covered by the Benchmark team was launched approximately 16 years ago. Admittedly it delivered monochrome images at 8fps, required a dedicated server and a single unit cost more than most mainstream systems. However, it signalled forthcoming change in the video sector. Today 360 degree cameras have become commodity items, but do they still deliver multiple benefits?
[dropcap]W[/dropcap]hen Philips CSI first unveiled the E-Dome in 1999, it was a radical departure from traditional CCTV camera design. The Benchmark team was amongst the first people outside of the manufacturer to see a working demonstration, and it was clear from the start that the technology offered a high degree of benefits.
Back then, the challenge was how to explain those benefits to installers and integrators, and indeed for engineers in turn to explain them to end users. This was because the camera changed the way that systems could be designed, installed and operated.
The simplest way to get across the potential offered by the camera was through a demonstration, but the E-Dome wasn’t the easiest camera to take to site for a proof-of-concept trial. It required a dedicated server for dewarping and image storage, and with a price that was often higher than an average mainstream system in its entirety, demo models weren’t too thick on the ground.
At exhibitions, the Philips CSI stand was easy to find; it generally had queues to see the E-Dome in action. Those in line included installers, integrators, end users, law enforcement agencies, consultants, specifiers and often a good number of competitors!
Whilst the E-Dome showed what was possible, it – along with the other 360 degree cameras that followed in its wake – failed to have a significant market impact. That had nothing to do with the cameras and their capabilities, but instead was due to the power (or lack of power) in computing at the time.
The cutting edge of PC technology in the late 1990s saw processing speeds of 500-600MHz, 128MB RAM, 12GB HDDs and 64MB video cards. If that sounds unimpressive, remember that to achieve such specifications required a significant investment. Average and more realistically priced machines featured 100-200MHz processing speeds, 32MB RAM and 1-4GB HDDs. Connectivity was also limited, and ISDN was the fastest (and fairly costly) widely available option.
Moore’s Law ensures that today we have more than enough processing and storage availability to run 360 degree cameras, and advanced chipsets allow high resolution captured images. The 360 degree camera has been pushed into the mainstream market primarily because advances in computing capacity allow full use of the devices.
It is important that installers and integrators consider how a 360 degree camera will be used, and what the customers’ expectations will be. There are two types of 360 degree camera: those which deploy a single image sensor – such as the models in this test – and those that use multiple image sensors. With single sensor models, typical use is to provide an overview image of a protected area.
Managing expectations for the cameras is important, because there are conflicting messages about the devices. All too often we see marketing materials making claims that a single-sensor 360 degree camera can replace numerous static or PTZ cameras in a protected space. The implication is that the 360 degree camera is a cost-saver, because it replaces numerous devices and their associated infrastructure.
Whilst such an approach might be possible in a few specific applications, in most it will result in the system not meeting an end user’s expectations. It might also mean the system cannot deliver the required and specified performance.
Experience has shown 360 degree cameras should be treated as an addition to a system, providing an overview of an area plus the ability to review footage with a variety of different and controllable views. The main objective is to ensure evidential continuity, and to allow an in-depth understanding of events.
This approach ensures that limitations with resolution and optical constraints do not restrict quality in the final image. It is worth noting that whilst many 360 degree cameras have megapixel resolutions, the total resolution encompasses the entire 360 degree field of view. When a usable region of interest is selected, the viewed resolution will only be a percentage of the total. Some cameras might further restrict resolution through the associated dewarping software.
Installers and integrators should also consider where the dewarping takes place. Some cameras will dewarp internally, which means you can only view and record selected parts of the image. Others will dewarp at the VMS or NVR, so ensure that compatibility is supported. Others transmit and record the full 360 degree image and use additional client software to dewarp.
The Benchmark test makes use of single-sensor 360 degree cameras, and considers their use as devices to supplement a larger video system with an overview of a protected area. The cameras were tested in an internal area, with ambient illumination that did not fall below 20 lux.
The DC-Y1513W is a networked 5 megapixel 360 degree camera. It utilises H.264 and M-JPEG compression and delivers a 2560 x 2048 pixel video stream at 30fps.
The device uses camera-side dewarping; this includes a full fish-eye view, single PTZ view, single panoramic view, double panoramic view, panoramic view with two PTZ views or main view with seven PTZ views. The selected video output is then recorded.
There is an option to use client side dewarping if either a DirectIP (the proprietary IDIS ‘plug-and-play’ platform) NVR or the company’s ISS software is used. The ISS software is licensed, but the ISS Compact version is available free-or-charge. ISS Compact is a single-server implementation of the software, which supports up to 32 IDIS cameras.
Our camera was shipped with a DVD containing a copy of ISS Compact. Unfortunately the DVD was not in a protective case, but had been put into a standard heavy duty envelope. As a result the disk was damaged and the software would not run. We visited the IDIS website and downloaded the latest version, but this did not support dewarping.
Benchmark contacted IDIS who confirmed that version 2.5.2 of ISS Compact is required to support the camera. This is now available from the IDIS website. The issue did result in a bit of downtime before we realised that the on-line software was out of date. We accept that disks do get damaged, but putting them into standard envelopes without any protection isn’t best practice.
The camera makes use of a 1/1.8 inch CMOS sensor, and sensitivity is quoted as 0.8 lux (F2.4). Other features include video motion detection, tamper protection, privacy masking, two-way audio, 90dB digital WDR, alarm I/Os and microSD-based edge recording. Power is PoE or 12V DC.
Physically the camera is typical of most single sensor 360 degree units. With regard to connections, the camera has a cable exiting the rear of the device with a flylead. This includes an RJ45 connection for LAN and PoE, alarm I/Os and audio I/Os using screw terminals, and a modular 12V DC power input.
The camera is supplied with a foam mounting pad, fixing screws and the afore-mentioned DVD. This contained the ISS Compact files which were unrecoverable, plus two PDFs: a camera manual and documentation for ISS Compact.
The Benchmark test did not use the DirectIP platform for the camera. However, if you do, discovery of and connection to the device is automatic. Outside of the DirectIP platform, connection to the camera is via a static address if DHCP isn’t used (typically it’s not used in security applications). Alternatively, the device can be found via ISS Compact. However, if it’s not on the same network segment you may need to reset it manually.
ISS Compact is beyond the scope of this test. However, it is fairly intuitive and connecting to the camera is straightforward. There is also an option to configure the device through the set-up module; typical operations are carried out via the client module. The set-up module has an option to effectively log you in to the camera webserver via the software rather than a browser. The process is simple and the menus are self-explanatory. There is a Basic menu, as well as more detailed menus for System, Network, Video, Audio, Event Action and Event.
With regard to camera set-up, there are options for sensor settings, white balance, exposure and day/night switching. The latter can be automatic, forced or scheduled. Streams can be defined regarding to camera-side dewarping (use the 360 degree image for client side dewarping), compression, resolution, frame rate, etc.. The panes – IDIS defines preset areas of interest as panes – can also be selected.
The camera also supports MAT: motion adaptive transmission. This decreases network load and required bit-rate by reducing frame rate during periods when no activity is detected. Frame rate returns to normal once activity is present in the scene. The function can be adjusted with regard to time of inactivity and detection sensitivity. There is an option to select base frame rate settings. It can also be disabled if not required.
The event menus allow motion detection, trip-zone and the alarm I/Os to be managed, along with sabotage detection elements. This does deliver a fair degree of flexibility with regard to alarm management.
Once configured, dewarping is carried out via the ISS Compact software. When the 360 degree image is called up, hovering over the frame displays a menu which includes dewarping. Basic dewarping displays can also be selected using a contextual menu.
Pan, tilt and zoom actions are carried out via mouse movements, although a joystick can also be supported. Pan and tilt motion is smooth, and the transitions are clean. It’ll take a few minutes to get used to the various options: both mouse buttons are used for different control options. Once you’re used to the process, navigating around live and recorded footage is straightforward.
With regard to image quality, the general overview is clean with accurate colour replication. Despite resolution not being as high as some of the other units on test, you’ll only notice the shortfall when zooming close on to objects or persons.
Dewarping is good. Processed video seems accurate, perspective is correct, and you only see signs of background processing when looking at edges close to the camera.
The ambient lighting in the test environment was relatively consistent. At its lowest it fell to 20 lux, and whilst the image was still usable, noise was evident and colour vibrancy was decreased.
The EVO-05NID is a networked 5 megapixel 360 degree camera. The camera is joint-branded between Pelco and Oncam Grandeye, although the hardware has more to do with the latter manufacturer. It makes use of Oncam Grandeye 360 degree imaging technology in conjunction with Pelco’s VMS and video management tools.
Using H.264 and M-JPEG compression, the camera delivers a 2144 x 1944 pixel video stream at 10fps. It can also deliver video streams of up to 1488 x 1360 pixels at 15fps.
There are a variety of viewing configurations. The camera streams a complete 360 degree view and uses client-side dewarping. This can either be managed via a compatible Pelco VMS or NVR, or Oncam Grandeye 3D dewarping software. The camera uses a 1/2.5 inch CMOS sensor and sensitivity is quoted as 0.2 lux (F2.0) for a 50IRE image.
Other features include regions of interest, video motion detection, privacy masking, two-way audio, digital WDR, alarm I/Os and SD-based edge recording. Power is PoE or 12V DC.
The camera is supplied with a quick start guide, a mounting template and that’s it. There are no included utilities or manuals, and if you want any supporting documentation you’ll need to visit the Oncam website and download it.
A brief look at the quick start guide indicates the use of a configuration tool. Again, despite being a necessary piece of software, Oncam expects you, the customer, to find and download this. You’ll actually end up with 11 versions, one for each supported language. The guide does have a URL for the tool, but it’s not correct!
There is an option to use a static address if DHCP is not running, or if you don’t want to download the utility yourself.
Making a connection to the camera for configuration is a bit hit-and-miss. On one server (optimised to allow ActiveX and other drivers) the config tool worked well. It found the camera immediately and allowed connection via a number of browsers. However, when the camera page loaded there was no prompt to load an ActiveX element, hence no displayed video, and the page layout (which makes use of concertina menus) did not display correctly. This made configuration impossible. On another server the config tool could not find the camera, but if the URL was entered manually then it would connect. However, again no ActiveX element was loaded. Instead it used Quicktime for display.
Whilst Quicktime works, it has high latency and suffers from frozen streams or missed frames. This is not just with the Oncam device; it happens with all cameras. As such, configuration is difficult to say the least.
To try and find a solution we applied a firmware upgrade. This did resolve the menu display issue, but the camera did not load its ActiveX element (as stated in the manual), instead defaulting to Quicktime.
The major issue here is that the hardware manufacturer, Oncam Global, has created a device but seemingly not worried about the configuration processes. Operation and dewarping is carried out via supported VMS and NVR solutions. Because Pelco fully supports the device is has co-branded the camera. However, the full configuration process might not be supported by all VMS and NVR systems, so we have to ask why the manufacturer has not provided an effective solution?
The camera module is covered with a plastic casing which is held in place with click-fit retaining lugs. Once removed this gives access to the various connections. These comprise network/PoE via an RJ45 socket, audio in via a 3.5mm jack, analogue video out via BNC, alarm I/Os and 12V power. There is additionally access to the memory card slot and two DIP switches for analogue output.
The set-up menus are limited to Image, Admin and Regions. Image covers basic viewing adjustments such as contrast and brightness, image sharpening, WDR (with a slider for level) and screen overlays. There is also a radio button selector to define camera orientation.
The Regions menu allows the creation of motion detection or privacy zones. These can include freely created shapes, and with motion a sensitivity level can be set.
The majority of the settings are carried out via the Admin menu. This has eight sub-menus, and an option to reset the camera to factory defaults. The sub-menus are Network, User Management, Alarm Settings, Camera Settings, Clock, Stream Settings, Audio and Recording. Most of the options are very basic. For example, camera settings are limited to resolution, compression, frame rate, bit rate (up to around 5Mbps) and GOP.
Alarm events are also quite basic, and consist of linking a trigger (VMD, input or network failure) with an action (Email, FTP, record to SD or output). Alarm durations can be specified.
If you want to familiarise yourself with dewarping prior to adding the camera to the system, there is a viewing tool but again you’ll have to download it yourself. It might not be worth doing so because in our experience it isn’t very stable.
Once in action, dewarping manipulation will depend upon the platform used, but it lacks the intuitive feel of some of the other cameras. Pan and tilting in the image always gives a slight feeling that the rendering engine is struggling, and the ‘feel’ of a dewarped view being standard – which many of the others do deliver – is somewhat lacking.
Image quality is also limited, which is unsurprising given that you’re viewing a section of an overall 4 megapixel image. Detail is average and colour fidelity is slightly off, with colours appearing muted even in decent lighting conditions.
In comparison to the other units in the test, the Pelco/Oncam Grandeye unit seemed somewhat out of date. It also gave the distinct impression of being a hardware device with little attention paid by Oncam Grandeye to the configuration process, coupled with an assumption by Pelco that configuration had been handled by Oncam Grandeye! The installer or integrator is one that loses out.
The IQS05FFACWCY is a networked 5 megapixel 360 degree camera. It is part of the Illustra Edge range, which is designed to allow the creation of a networked video solution without a need for additional servers. It operates using Tyco’s exacqVision Edge platform.
The camera utilises H.264 and M-JPEG compression, and delivers a 1936 x 1936 pixel video stream at 14fps. The camera streams the 360 degree view and uses client-side dewarping via the edge version of the exacqVision VMS.
Using a 1/2.5 inch CMOS sensor, sensitivity is quoted as 1.2 lux (F2.0). Other features include digital noise reduction, WDR, motion detection, privacy masking and microSD-based edge recording. Power is PoE.
Because the camera comes with an edge VMS solution, the set-up is a little different to many of the other cameras. The edge package can be upgraded to a full version of the Exacq VMS, and this is an open platform system so can also be used with many third party devices.
Whilst exacqVision is beyond the scope of this test, Benchmark has looked at it in the past and it’s a stable and usable platform, albeit without some of the advanced functionality included in other VMS options.
One word of warning; the operating software makes use of the supplied and fitted microSD card. During set-up the protruding edge of the card was caught, which was enough to release it. It was replaced straight away but the system wasn’t having any of it. In the end we had to power it off and start again!
Once the camera is connected (LAN/PoE connectivity is via a rear-mounted RJ45 socket), you can verify the settings via Internet Explorer if required. It’s worth noting that if you don’t want to use the Exacq edge software you can configure the camera via a browser, but this process uses Quicktime. As mentioned previously, Quicktime isn’t great with high resolution streaming video, and can make for a slow and clumsy set-up.
We opted to use the Exacq edge software. This quickly finds the camera and also tests if you’ve read the manual! The camera log-in is different if you’re accessing it via exacqVision. With the right authorisation, the software automatically creates all the required settings for a basic start, and from there onwards it’s a matter of tweaking the settings. You can also access the full range of camera configurations via the software.
The VMS interface uses a Windows-type tree, and from this you need to select the device and expand the options. This reveals icons for System, Camera Recording, Video Input and Audio Input/Output. Selecting each brings up a series of menus to allow full customisation.
Much of the set-up is straightforward, and even those new to the Tyco VMS will find it intuitive. There is full documentation, albeit on a DVD in PDF format.
Video settings include content adjustment (contrast, brightness, etc.), WDR (off or auto), recording quality, motion detection, PTZ control and dewarping, and schedule. There isn’t anything that should challenge a competent installer or integrator.
Viewing options are fisheye view disabled (oddly, this gives the fisheye view without any manipulation options), dewarp (this gives a single view with ePTZ functionality), panorama or double view (two panoramic 180 degree views). With ePTZ functionality, control is smooth and dewarping is accurate.
In truth, the panoramic view is shown in its entirety which makes it next to useless in all but a few applications.
When viewed alongside the other devices in the test, the Achilles’ heel of the IQS05FFACWCY is image quality. Although the unit is billed as a 5 megapixel units due to its chipset, the delivered video is around 3.7 megapixels
for the entire fisheye view. Any region of interest is subsequently going to only include a percentage of that resolution, and by the time you consider the impact of processing and bitrate management, the image is only suitable as a general overview.
The video quality has 10 levels, and adjusting these didn’t really have an impact in as far as the viewed image improved or deteriorated. In fact, the only way of noticing that the changes had any effect was to look at the used bitrate if VBR was selected.
Identification of individuals was possible, but if required for evidential purposes a video stream from a higher quality device would be required to back-up the continuity offered by the 360 degree unit. The degradation was also problematic if the identification of details, such as lettering on an object, needed to be captured. Using digital zoom simply increased the problem.
Ambient lighting in the test environment was consistently above 20 lux, and as the light levels fell the camera coped well. However, we’d have happily sacrificed this element of the camera’s performance for more image detail.
The CBP360-IP is a networked 6 megapixel 360 degree camera. The camera utilises H.264 and M-JPEG compression, and delivers a 3072 x 2048 pixel video stream at 30fps. The camera streams the 360 degree view and uses camera-side dewarping; it can be client-side too via a compatible VMS or NVR.
Using a 1/1.8 inch CMOS sensor, sensitivity is quoted as 0.2 lux. Other features include integral infrared illuminators with a stated range of 10 metres, digital noise reduction, WDR, highlight compensation, audio connectivity and microSD-based edge recording. Power is PoE or 12V DC/24V AC.
Connections are made via a flylead. This includes network/PoE via an RJ45 socket, audio out, alarm I/Os and 12V power. Access to the microSD slot is via a small cover beneath the fascia ring on the main unit.
The camera is supplied with a quick start guide, connectors and a miniature CD. The latter contains the company’s Network Video Manager System as well as full documentation and other tools. The camera is also supplied with a paper quick start guide which covers basic installation and set-up. For those who have used Concept Pro devices before, the format will be familiar; if you haven’t then the instructions are concise but clear and easy to follow.
When it comes to initial configuration, you have three main options. The first is to add the device to a network with DHCP enabled, allow an address to be allocated and then look this up at the router. The second is to use the NVMS software and the third is to add the device to a non-DHCP network and use its default static address. The NVMS option works well; it finds the camera quickly, allows a connection to be made, and within a very short time the camera is running.
On initial log-in via the browser, you are prompted to download the latest version of Flash Player, as the interface uses the plug-in. There is an option via a link to use ActiveX instead. While manufacturers seem to be moving away from ActiveX, we think it’s better than Flash, plus if you want to set up motion detection you’ll need it as that function only supports ActiveX views to set zones!
Camera configurations can be adjusted via the software, but we couldn’t find a way to adjust the streamed content, and received a default view with the full fisheye plus five PTZ views. It was then a case of using a browser to select the fisheye streams. Options are full 360 degree fisheye view, fisheye plus three PTZ views, fisheye plus five PTZ views and fisheye plus seven PTZ views.
Selecting the PTZ views is straightforward. Highlighting a given view shows the region on the overall fisheye view. This can then be dragged to the relevant position. Sizing of the views cannot be adjusted. It’s a slight limitation, but you have to remember that this device is firmly aimed at the cost-conscious market. On the up-side, it does help to retain a good level of video quality.
Camera stream configurations include format (H.264 high, main or base, or M-JPEG), resolution (3072 x 2048, 1920 x 1920, 1920 x 1080 or 1440 x 1440), frame rate (up to 20fps) and bit rate (up to 12Mbps). The core configurations can be adjusted via NVMS or direct from the browser.
Alarm events are simple to configure, but in all honesty they are basic. In systems using a VMS you’ll be better off using the software’s alarm handling funtionality.
The dewarping is clean and consistent without any issues rendering perspective. The control will vary dependent upon the interface. One good point is that Videcon provides notes for a range of supported VMS interfaces. The camera does lack some of the more flexible functionality of the others.
Image quality is good considering its resolution, and degradation only puts in an appearance when used with digital zoom. A drag-and-click function for zooming does mean that at times you overdo things, but it is a matter of understanding how far you can push the function.
The ambient lighting in the test environment was consistently above 20 lux, and there were slight signs of noise when it fell to the lowest levels. However, this did not make the image unusable at any point.
Whilst beyond the scope of this test, the camera’s infrared illuminators work well and deliver consistent coverage.
The benefits of 360 degree cameras are obvious, and the units undoubtedly can enhance the overall security of many applications.
The DC-Y1513W from IDIS is a good performer, but it has been designed for use in an IDIS system. If you’re happy to go down that route, then it’s a credible add-on. However, if the open platform approach is your chosen path, then on-camera dewarping does lack flexibility. The camera is recommended with the proviso that installers and integrators are comfortable opting for a one-brand solution.
The EVO-05NID from Pelco is actually an Oncam Grandeye unit and is joint-branded. It is worth noting that Oncam makes the camera and Pelco offers compatibility via its VMS and NVRs. The test is of the camera itself. Compared to the others it is underwhelming. Configuration is difficult because the utilities and tools – all of which you must download yourself – aren’t stable and lack the quality of other devices. Image quality is average, and performance falls behind that of the competition.
Tyco’s IQS05FFACWCY is an interesting proposition. The camera is supplied with a free edge version of the company’s exacqVision VMS, and as this is an open platform VMS you’re not tied in to Tyco products if you want to upgrade. The camera is also supplied with documentation, PDF manuals and utilities. Installation is simple and dewarping is basic but very stable. However, the video stream gives a 3.7MP fisheye view, and once you start dewarping you soon realise that video quality is only good enough for a general overview. No one expects high quality from these cameras, but the IQS05FFACWCY struggles in this department.
The CBP360-IP from Videcon represents a 360 degree camera for the budget-conscious. It lacks the depth of control of some of the leading units, but does offer enough flexibility for many mainstream applications, dependent upon the supporting VMS selected. Image quality is retained as views are kept relatively small.
The Best Buy is the Panasonic WV-SFN480 (tested last issue in Part 1). It delivered in terms of functionality and stability, and while it didn’t have the highest resolution, image quality was very good. The features allowed a good degree of flexibility, and performance was stable.