All posts by lindanatale

I Can See Clearly Now: Fisheye Cameras and Dewarping

There’s much to like about 360-degree fisheye cameras.

Their ability to capture images that could be missed by a traditional fixed camera or even a pan-tilt-zoom only model makes them critical for security applications where sweeping views are required. For example, 360-cameras excel within settings like an open concept retail store where a single camera can be observing all directions at the same time. The cameras have also proven useful beyond the retail setting in college lecture halls, casinos, office lobbies, hallway intersections and enclosed parking garages, to note a few.

Fisheyes are also touted for their wide dynamic range, as well as savings on bandwidth and storage usage while still capturing full resolution images.

The image produced isn’t a typical, flat one, but rather is round and distorted, the result of capturing an ultra-wide field of view. To make the images usable, they need to be flattened out, or dewarped. In most cases, the dewarping process takes place at the client, such as the NVR, PC, or server depending on where the client software that handles the dewarping of the images resides.

Fisheye Cameras DewarpingA few manufacturers offer cameras that can natively dewarp the images. Thus, the video can be viewed in the more traditional format, but with the benefit of getting all the detail that a 360-degree view provides. Having the camera perform the dewarping process relieves the burden and reliance for this action from the recorder and client, bypassing the step of having to dewarp the recorded view separately.

In dewarped or flattened images, security personnel can hone in on details that are important to the situation, while also having the option of looking at the larger picture. If you think about a 360-degree image as four quadrants — north, south, east and west — consider that when an event occurs, you have the option of retrieving an image that shows action in the north quadrant only, where you believe the suspect activity occurred.  There may be something that took place elsewhere that you now want to check out, and so you can playback images that capture activity in the south, east and west quadrants as well.  For this reason, fisheye cameras are a cost savings as they can take the place of four cameras when placed at the intersection of hallways.

Having quick access to a normalized view of a situation, especially when an event is occurring, makes dewarping at the camera level a convenient tool for seeing everything the video has to offer.

 

In what ways are you using Fisheye Cameras? Please leave me a comment below.

No more one size fits all – welcome to vertical view

Like the round peg trying unsuccessfully to fit into the square hole, many surveillance cameras have been similarly hampered by trying to reconcile their horizontal nature within a vertical video monitoring scenario.

Vertical ViewConsider all of the security-related situations in which a vertical camera image would be preferable over a horizontal one — policing the long, tall aisles in a big box store, home improvement center or supermarket; or watching over the straight, narrow corridors found in hospitals, schools, and even some office buildings.

In all of these instances, a vertical or portrait view of the scene would be preferable to a horizontal one. So rather than lock the user into a 16×9 horizontal world, if the user needs a full HD quality video stream, why not allow the maximum number of pixels to the field-of-view (FOV)?

Today most security cameras have been designed for the horizontal perspective so simply rotating the camera to a 9×16 FOV sounds easy enough, but the resulting video is simply a sideways oriented 16×9.  In addition to rotating the camera or effectively turning the imaging sensor on its side, the data recorded by each of the pixels on the sensor has to be rotated to reorient the video stream into a true 9×16 streaming HD video. Sounds simple, but doing so without quality loss or sacrificing video frame rate can be a challenge.

In theory, of course, it is possible to cover those narrow, vertical corridors and aisles with traditional landscape-style camera views, but it will take additional cameras — and additional cost — to achieve it. This is not only limited to camera and installation costs, because each of the cameras deployed will be recording additional scene area to the left and right of the center scene, which adds cost for bandwidth and storage to the overall cost calculation.

Fortunately, the industry is now responding with highly efficient mini-bullet and mini-dome cameras that have the higher-level processor and memory capacity so the camera can be placed in permanent portrait mode. It’s not just about switching the aspect ratio from 16:9 to 9:16 for these cameras, but it’s also about having sufficient horsepower to properly achieve and support this functionality with no residual impact on the camera’s resolution or frame rate performance.

By having cameras that reconcile properly with the space they are covering, operators will get the most complete, most usable images for active surveillance and forensic purposes. No more wondering what has been cut off from the picture, no more parsing together images to get a complete look.

And when these new cameras are made part of a larger system that includes an NVR with preconfigured layouts to accommodate the taller, narrow viewpoint, it becomes an even more ideal fit for viewing, archiving, and retrieving these images.

Kind of like putting that square peg into the square hole. It’s a perfect fit.

 

What other ways are/could you use a vertical view as part of your security solution? Let me know in the comments section below.