Before we move on to the actual working of Augmented Reality, we first need to know what it really means.
Augmented Reality is basically a mix of both the virtual and the real world, an interactive enhancement of the real world, by means of superposition or overlaying of Digital objects into our real physical world. It is basically a blend of our real-world environment with some virtual objects placed in it, in a three-dimensional manner. Most case, these virtual objects can be controlled and viewed by us, in the same way when some real-world objects placed right in front of us.
For example, when we play Pokémon Go and we see virtual images of the Pokémon in our real world.
The working of Augmented Reality devices is based on the principles of image registration and Simultaneous localization and mapping (SLAM).
How does Augmented Reality work??
Image Registration:
Image Registration is the estimation of distance depth and coordinates of the real-world entity, through studying the camera images.
This principle comprises of the following steps:
Identifying interest points or markers (Geometric Registration):
Images of physical objects are transferred from the camera to the AR device. Certain regions called fiducial markers are marked on the image of the object. The device mainly consists of hardware like a gyroscope, GPS, digital compass, accelerometer, etc.
The function of a gyroscope is to measure the angle and orientation of the device, GPS and compass get direction and location of the device and the accelerometer is used to detect relative motion and speed of the object.
All these parameters are measured with respect to the markers on the images keeping them as a reference point.
- Acquiring exact orientation of physical entity:
Now that we have parameters of distance, depth, speed, orientation, etc. of the device, the device employees visual route at this point.
Through this technology, the device can re-establish the real-world coordinate system of the physical object using its image and the data acquired from it.
In other words, the device now knows exactly how the physical object is positioned and oriented, in the real world three-dimensional space
2. Mapping the virtual object to real environment (Point Set Registration):
Now we have the real-world coordinates of the physical entity. The device is ready to examine the virtual 3D object into these coordinates in such a way that the virtual object seems to be in perfect orientation with the real world 3D coordinate system. This usually refers to computer vision and pattern recognition, also known as Point matching.
This overlaying of the virtual object into the real environment is bascically achieved through a technique called point set registration.
In this the two point sets are, the set of predetermined points on the virtual object, and the group of interest points on the real .
Point set registration in Augmented Reality refers to a process in which the points of the virtual objects are mapped to the fiducial markers present on the physical entity. This further results in the formation of an Augmented Reality model in which a virtual object is perfectly aligned positioned and oriented within the real world three dimensional space.
Augmented Reality – Location-related Advertising / Virtual based Information
The future of location-based virtual display information has various predictions from sci-fi/horror hidden AR messages in They Live, to a lighter vision of Augmented Reality hyper advertising from Keiichi Matsuda. Both of these film narratives portray virtual information displayed all around are based on our geolocation.
Mobile geofencing, and it’s just a matter of time until next-generation GPS is available to allow for AR-based objects to be placed with precision in (virtual) location.
Though this technology is similar to how HUD real-time display information will be displayed, there are misleading technical challenges with either approach.
Whereas automotive-based HUD’s can utilize more powerful imaging devices and processing power for object detection/recognition, personal eyewear will be subject to less processing power and will have to rely on using geolocation to display virtual object or any other details. It will be a while before a wearable device can do accurate, distance-related object detection/recognition.
Use of Augmented reality in Emergency management/search and rescue
Augmented reality systems are used in public safety situations, from super storms to suspects at large.
The first was “Augmented Reality–Emerging Technology for Emergency Management” by Gerald Baron. Per Adam Crowe: “Technologies like augmented reality (ex: Google Glass) and the growing expectation of the public will continue to force professional emergency managers to radically shift when, where, and how technology is deployed before, during, and after disasters.”
Another early example was a search aircraft looking for a lost hiker in rugged mountain terrain. Augmented reality systems provided aerial camera operators with a geographic view and awareness of forest road names and locations which are blended with the camera video. The camera operator was able to search for the hiker knowing the geographic context of the camera image. Once located, the operator could more efficiently direct rescuers to the hiker’s location because the geographical position and reference landmarks were clearly labelled. So AR basically helps to locate from a tiny object to a person lost in rough terrain places or difficult places
An AR mobile game using a trigger image as fiducial marker
Non-rigid Point set Registration Methods
Basically what happens here is that GMM centroids are uniformly positioned along the contour which allows for non-rigid transformations. In reality this method is only applicable contour like point sets. One of the most popular non-rigid point set registration methods is by Chui and RANGARAJAN.
WRITTEN BY -ARYAMAAN BANERJEE
REFERENCES
The Technical Prophet 