Design for 360 Degree Camera

Camera enclosures for outdoor use require associated mechanical pan and tilt equipment if the camera is needed for anything further than just general area coverage. Not only do these mechanical devices take critical seconds to slew to position, they also require regular maintenance. What if the pan and tilt functions could be done purely electronically, and thus in near real time—using the same camera, itself in a fixed stationary position?

This could be done with a convex parabolic reflector (PR) that is oriented perpendicular to the field of observation. For example, if the objective is to monitor a parking lot, the PR would be pointed directly at the ground. The camera would be positioned directly under the PR, and pointing directly up at it. The entire assembly would be mounted on a pole of a height selected to provide optimal field of view and lateral orientation detail. The PR and camera assembly could also be oriented horizontally, for example to provide a view of aircraft both in the air and on the ground, or of a wildlife preserve, oceanfront, or the side of a mountain, building, or other tall structure.

Ideally the PR would consist of polished gold plate, tempered glass mirror, or other durable material that would not lose its reflectivity due to oxidation or other corroding effects. Naturally the PR and camera lens would need to be dusted off occasionally, which would be possible to do by any number of simple means.

The camera’s lens would necessarily be optically calibrated to cover the PR’s surface pointing at it. The lens would also need to compensate for the distortion of the image coming off of the PR, especially toward the edge of the field of view. This optical calibration calculation would need to be done only once per match between lenses and PRs of different sizes, as there would be no need to change the physical assembly after it’s constructed. All operation after installation is by electronic means.

With the lens properly calibrated, the camera’s sensor receives a flat image that captures the full surface of the PR. Software would then select portions of the image to display with virtual pan and tilt functions, none of which need any physical movement at the PR/camera assembly. As the image presented to the camera will be a flattened version of something similar to looking down at the earth from directly over one of the poles, the software would also need to rotate the selected image area the number of degrees necessary for it to appear right side up when displayed—something trivially done in most elementary image processing utilities, and which should be readily doable with a reference to the general coordinates within the main image the selection is drawn from.

As the camera sensor is constantly receiving an image covering the entire (virtual) pan and tilt range, custom software alerts could be configured for anything ranging from motion detection to scheduled rounds of pre-programmed rotating images from virtual slew coordinates.

An optional enhancement would include addition of a thermal (8-12μm) image sensor in addition to the normal visible and near infrared light senor. A composite lens might be used, consisting of a larger germanium lens for thermal with a silicon lens in its center core. This would provide identically aligned optics in both wavelength ranges, to provide the same wide angle views for both visible and thermal images. The respective sensors would be selectable in the software, which could additionally analyze all collected data together.

All of these capabilities are made possible by use of a convex parabolic reflector paired with a corresponding camera aligned to its axis of symmetry, a lens optically calibrated to provide an undistorted flat image to the camera’s sensor, and associated software for virtual pan and tilt functionality.