The growth in demand for panoramic cameras has seen offerings available at virtually every level, ranging from high-end top specification units down to simple plug-and-play budget cameras with integral dewarping. With most manufacturers now offering a 360 degree solution, there is still some confusion over the best practices when deploying these devices. Benchmark considers the role of panoramic cameras to assess how they can be best used.
Panoramic video offers a great many benefits in video surveillance applications. It can provide an overview of an area, enabling all parts of a protected area to be recorded. This enables users to retrospectively pan, tilt and zoom in recorded streams to assess incidents. Panoramic video can also allow the creation of visually corrected wide angle images, showing continuity that is difficult to achieve with multiple cameras. Increasingly, panoramic cameras can be deployed to automate PTZ domes cameras.
Panoramic devices have been available to system integrators and end users for many years, and it that time the technologies used to create 360 degree images, and the operational requirements of those seeking to deploy the devices, have changed significantly. It is no longer sensible for integrators and users designing and delivering smarter surveillance solutions to consider all cameras which can deliver a 360 degree image in the same way.
Advances in image sensors, lens technologies and – importantly – video processing have allowed manufacturers to approach the delivery of panoramic video footage in varied ways. For integrators and users, this ensures a selection of technologies which can be specified dependent upon users’ needs.
Today 360 degree surveillance can be achieved in a number of ways. Either a dedicated panoramic camera with a fisheye lens can be specified, a standard high resolution camera can be fitted with a panomorphic lens, or a number of static cameras can have their images ‘stitched’ together to create what is viewed and managed as one single panoramic view. The choice of which approach to follow is very much dependent upon the requirements of any given site.
Panoramic cameras can deliver all-round 360 degree views or wide angle 180 degree views. Some manufacturers will describe the latter options as hemispheric cameras. Less common are 270 degree cameras; these are designed to be mounted on an external corner of a building, allowing for all open spaces to be covered.
Considering implementations
The main approaches when creating a panoramic video stream do have some impact on how the video can be best used. The first (and most common) method makes use of a single sensor (or single standard camera) fitted with a fisheye lens or an after-market panomorphic lens.
This approach is widely used by mainstream camera manufacturers. Many will often a selection of different resolutions with these devices. These cameras will capture a distorted fisheye view, and the image is then corrected via a dewarping process. Dependent upon the model in question, dewarping either happens in the camera itself or at the VMS or NVR.
The second approach involves the use of multiple camera sensors, typically four with each capturing a 90 degree view. These different views are then digitally corrected and stitched together to create a full 360 degree view which is treated as a single camera view. Again, the processing can occur in the camera or, more commonly, at the VMS or NVR.
In both cases, once the raw video footage is captured, it undergoes some fairly intense processing to deliver a variety of different displays and management options, again dependent upon the applications’ needs.
While both approaches deliver panoramic video streams, there are important differences in how the final video stream can be used. This involves the resolution of the video and impacts on display rather than image capture.
With single sensor cameras utilising a fisheye lens, the camera captures a full image of the entire 360 degree area. In its raw format, the video is pretty much unusable. However, by using a dewarping algorithm, software is able to adjust perspective in the video, thereby presenting a visually correct image.
When considering the resolution of single sensor fisheye cameras, the quoted figures are for the entire image. Therefore, if a camera is specified as delivering a 5 megapixel video stream, the displayed corrected image will not be 5 megapixel. The quoted resolution will be for the full fisheye image.
It must be remembered that the captured image will not fill the full chip surface, as a fisheye lens delivers an elliptical image. This means that some of the available resolution is lost before any calculations being. To ensure the full panoramic image can be viewed, the corners of the sensor do not capture usable video. Therefore, if a 90 degree view is displayed, dewarped from the full captured stream, the display will typically have less than 25 per cent of the quoted resolution. If any degree of digital zooming is carried out, the displayed resolution will subsequently decrease.
It is also important to realise that quoted display resolutions may allow for upscaled images. This approach is used to ensure that displays are native for monitors, and is not an indication of video quality.
With multiple sensor 360 degree cameras, integrators and users can be a bit more assured about the final image quality. For example, if a multi-sensor devices makes use four cameras and each has a resolution of HD1080p, the total resolution may well be quoted as 4K UHD. This equates to all four 90 degree elements of the 360 degree image being displayed as a quad view.
Of course, the camera will allow the images to be displayed in a number of different ways such as 2×2 (4K UHD) or a stitched 360 degree view (7680 x 1080 pixels).
Because each of the multiple image sensors has a specified resolution, it allows integrators and users to be certain that a 90 degree region of interest will have an exact image resolution: in the case of this example, HD1080p. Obviously this will reduce if digital zoom is deployed, but only in the same way that it will for a standard HD1080p camera.
It is important for integrators and users to calculate the image resolution of any dewarped and displayed scene, as this will be the difference between a camera suitable for providing an overview of a protected area, or allowing more detailed analysis of activity and incidents in the camera’s field of view.
A flexible option
Panoramic cameras allow video to be displayed in a number of formats. Typically these include 360 and 180 degree panoramic views, as well as custom regions of interest which can be viewed in various formats including 4:3 and 16:9 aspect ratios.
Because the cameras stream the entire 360 degree image to the recording server, this means that all details in the viewed scene are recorded. Therefore, an operator has the ability to view any part of the protected area when reviewing the recordings. This allows them to rerun the video and view different areas, depending upon their needs.
Some budget-level 360 degree cameras perform dewarping at the camera. This, however, defeats the purpose of these devices, as the recording server will only capture the dewarped streams. If an operator needs to change the field of view when assessing recorded video, they will be unable to do so unless the full 360 degree image has been archived.
With multi-sensor devices, the individual cameras capture traditionally formatted video so there is less requirement for dewarping. The processing will often involve some correction of barrelling where very wide angle lenses are deployed, and to stitching of individual streams into one 260 degree image. These devices operate as a single camera, so there will be a single network connection and a single power connection (unless PoE is used).
With regard to video display, either the multiple camera outputs can be viewed alone, as a quad display, or as a panoramic view. With the latter option, it is then possible to electronically pan, tilt and zoom within the panoramic image without any need to change views.
Realistic expectations
The benefits of panoramic cameras are obvious, and the units undoubtedly can enhance the overall security of many applications. When installed correctly, and when considered with realistic expectations, these devices represent exceptional value for money. Despite this, there does remain a degree of negativity from some in the industry towards these units.
Often, unrealistic expectations for what can be achieved exist, with confusion over the expected levels of image quality and suitability for certain tasks.
It is not uncommon to see some panoramic cameras promoted as being a cost-effective alternative to numerous static devices. This message is not unusual, but in the vast majority of cases it is not accurate.
Whilst there are some high resolution exceptions to the rule, many panoramic cameras – especially single sensor models – are designed to provide an overall view of the scene. They are designed to be used for general surveillance in addition to other static or PTZ cameras, rather than as a replacement for all other devices.
As already mentioned, it should be remembered that in a dewarped scene, the viewed area will represent a small percentage of the total image and subsequently of the total resolution. Add digital zoom and the quality can quickly deteriorate.
Interestingly, a number of credible manufacturers who specialise in panoramic devices (and have many years of experience in this field) often advise that single sensor panoramic cameras only be used to provide an overview. The video allows general activity to be tracked and continuity to be proven, but may not always deliver images suitable for evidential purposes.
Resolutions are increasing, but it still remains the case that in the majority of applications, the cameras won’t be suitable to replace multiple dedicated devices. Indeed, the strength of panoramic cameras is not to replace multiple static units.
A smarter option
A smart approach can be taken with panoramic cameras if deployed using intelligent video analytics or advanced motion detection. When an alarm condition or movement is detected, the resultant action will send a linked PTZ camera to a pre-set position covering the triggered detection zone. As this approach allows the use of optical zoom on a fully functioned camera, image quality is preserved, even over long distances.
The combination of a panoramic camera coupled with a fully functioned dome delivers both evidential quality video for identification and general wide-area surveillance to provide evidence of continuity.
The true potential of panoramic cameras comes from the elimination of blind spots, coupled with the ability to alter the camera view in recordings.
In summary
Panoramic cameras may have had their true benefits misrepresented in the past. The reality is that good quality panoramic cameras won’t reduce the cost of an installation. They still represent good value for money, because the benefits can enhance site security to a very significant degree.