Augmented reality has come a long way since we first talked about it two years ago. Though the technology itself remains largely in its experimental phase, its potential for future applications of use is tremendous. Recently, Sony has been working on abit of AR themselves and they call it SmartAR.
This technology enables additional information to be displayed on Sony’s device screens such as virtual objects, or images and texts that cannot be identified by visual perception alone (see video below).
The technology employs the markerless approach, which means users won’t need to use any special markers such as 2D barcodes. The object captured by the camera is quickly recognised and is then tracked at high-speed along with the movement of the camera, as it is displayed over the actual 3D space.
See official full press release below…
AR information can be displayed on the captured image which appears on a device’s screen, including those images that do not have any special markers for AR. This technology is also compatible with image recognition technologies that use conventional markers (such as “CyberCode*2”). Because “SmartAR” can recognize everyday objects such as posters and menus, it has the potential for a wide variety of applications. “SmartAR” object recognition technology identifies objects by analyzing features detected from a portion of the image (hereafter, ‘local features’) together with their positional relationship. Our feature matching technology that employs a proprietary probabilistic method that matches local features with minimal calculations enables high-speed recognition that is resistant to changes in lighting or the position of the object. In addition, recognition is still possible even if the object captured appears to be comparatively small in the display.
High-speed tracking (‘rapid & accurate’)
Sony achieved its natural-feeling ‘rapid & accurate’ AR by quickly displaying AR information on the screen and then tracking the camera’s movements at high-speed. This has been realized by combining object recognition technology with our proprietary matching technology that uses features detected from a portion of the image (‘local features’) and image tracking technology that is capable of dealing with changes in the shape of the object.
3D space recognition
With our dynamic, large-scale AR, virtual objects can be merged with 3D structures detected in the physical world. For example, even if the AR image is a gigantic virtual character whose size exceeds the dimensions of the device’s screen, the technology allows the user to grasp the entirety of the virtual character when the camera is moved around. Furthermore, it is also possible to move the virtual object in the actual 3D space as if it were really there.
Three-dimensional space recognition technology is based on use of the disparity observed by the camera movements to estimate the shape of the 3D space and the position and angle of the camera. By combining this with object recognition technology, devices become capable of identifying and remembering 3D space constructions.
Information can be intuitively acquired and navigated by directly touching the AR information displayed on the smartphone or device’s screen. The distinctiveness of “SmartAR” technology comes from the user interface which enables users to naturally use and operate additional information and virtual objects.