In this study, based on AR research related to realistic,
hands-free and wearable, wide field of view, and collaborationfriendly devices, an interior design system was implemented.
In the early stage of projection-based AR studies for designs,
the studies initially projected new textures, colors, and patterns
onto objects. Gradually, as the technology developed, it was
not limited to projection on objects, but expanded to spaces and
could be used in various fields. However, these systems simply
provide visual feedback to the user and do not offer
interactions with objects. In other systems, users can use a
controller with an infrared sensor to actively participate in the
designing process by projecting on pre-selected objects [10],
[11]. This direct designing on objects offers the user a high
level of accomplishment and satisfaction with the DIY process;
however, since the interaction is possible only through the
controller, it can be inconvenient for the user.
The projection-based AR studies that extended from
object-oriented design to space-oriented design were initially
carried out to apply digital textures, colors, and patterns on
buildings and throughout walls during the early stages of media
arts [12], [13]. However, the early studies of space-oriented
designs, similar to earlier studies of object-oriented designs,
provided only pre-set information for projections. Since then,
studies were carried out that involved space designs with
several projectors installed in a space with reconstructed
arbitrary walls, rather than projecting the previously set
information [14]. In the case of this study, user interactions,
such as painting arbitrary walls with colors using a controller,
are enabled. In addition to this study, by installing several pantilt-based projector-camera units on a ceiling and projecting
pre-designed room interiors onto the actual space in real life,
another space design study was carried out to construct a
highly realistic AR space [15]. The study provided users with
various cases of highly realistic interior designs. Also, the
virtual effect can change if the user moves around in
constructed space. In another study that designed a highly
immersive virtual space, the study was conducted in an
experimental environment, similar to the study mentioned
earlier, where several pan-tilt-based projector-camera units
were fixed on the ceiling [16]. Gestures and body interactions
with a projected space are possible, but the interface for such a
space-designing purpose is non-existent, since current
interfaces only provide predefined visual feedback based on
user activities. Since designs, such as textures and patterns,
applied to objects and spaces using a projector add a touch of
realism, design-related studies are actively ongoing, even
though most space design studies involve a lot of time and
money due to the fact that multiple projectors must be installed.
In this paper, we provide users the experience of designing
interiors in their own personal space via the DesignAR system,
the only one constructed specifically for the purpose of interior
design prototyping. Since it is manufactured as a portable pantilt-based projector-camera system instead of a ceilingmounted system, the system can be simply and conveniently
used in various environments. A spatial user interface for
interior design is also constructed, and by letting it link with
mobile devices, efficient interior design is now possible in
various situations.
http://www.aleeshainstitute.com/interior-designing-course.php
The portable pan-tilt mechanism of the DesignAR system
allows for simple installation and wide space coverage with
only one system. The interior design of the actual environment
via projection requires information about the space to be
designed, which can be effectively performed by space
information construction. Also, it is necessary to reduce the
complexity of using the system by providing only the space
information required by the user. For this reason, the
DesignAR system constructs and analyzes space information
for interior design as 3D information. With a single pan
rotation of 360 degrees, DesignAR constructs a 3D map of the
space and objects in its surroundings. By constructing such a
3D map, DesignAR provides users with real-time information
about the surrounding space, and then, based on the
constructed 3D map, it recommends possible planes that the
user can apply in their interior designs.
A. Hardware configuration
As shown in Fig. 2, in order to provide users with an
interior design interface, we created a projector-camera unit
system with a portable pan-tilt mechanism. It is designed to
enable a pan-rotation of 360 degrees and tilt–rotation of 180
degrees through two mounted servo motors.
This system conveniently allows the user to project onto a
desired space, regardless of the position of the portable pan /
tilt system. On the pan-tilt system, a projector is installed to
provide realistic visual information to the user. The RGB-D
camera installed on the projector allows the system to collect
information regarding the surrounding space and user
interactions. Users can set the desired angle by controlling two
servo motors with Arduino.
B. Space information processing for interior design
The information about the space is received via the
projection-camera unit, and through the construction of the
space, information can be delivered to users in the form of 3D
space data. In addition, by analyzing the space, the optimal
space for interior design prototyping can be offered to users,
thus improving the system’s convenience. The study applied
the DesignAR system as shown in Fig. 3.
1) Space construction: In order to construct the user's
surrounding space, the feature points are extracted using depth
images and color images; the panning servo motor is then
rotated at a pre-determined angle to obtain these images via an
RGB-D camera. In this paper, the angle was 5 degrees per
rotation. The feature points of the input image are then
extracted by using FAST [17] and BRISK [18] to perform
matches between frames. Frames are matched using 2D
matches based on color images. After that, to calculate the
final pose, reconstruction is performed, based on the 3D space,
using the depth images. With the 3D map constructed this way,
information about the surrounding space can be provided to
the user.
2) Space analysis: Through space analysis, optimal planar
space is recommended to users; in this paper, a Modified
RANSAC algorithm [19] was employed to find the optimal
planar space. Compared with the RANSAC algorithm that
performs random sampling, the Modified RANSAC algorithm
selects the sample data by down-sampling, and this improves
the speed and detection performance. The Modified RANSAC
algorithm is composed roughly of two processing steps.
During the first step, the hypothesis step, in order to generate a
model that satisfies the sample data, three points are randomly
selected at a certain distance from the point detected. In the
second step, the verification step, the planar model is
evaluated after the extraction of inliers. When there are more
inliers than in the existing plane, the optimal plane is detected
by updating with a new model.
Comments
Post a Comment