IEEE Consumer Electronics Magazine - May 2018 - 76

Table 3. The parameters of a transverse
Mercator map projection.
Name

Symbol

Value

Equatorial radius

6,378.2064 km

Flattening factor

1/294.9786982

Latitude of origin

0°

Central meridian

79.5°

Scale factor

0.9999

False easting

304.8 km

False northing

0 km

FIGURE 3. A visualized Toronto 3-D city model (without city data)
on a gray plane using HoloLens.

FIGURE 4. The interactions between a user and the augmented
contents.

divided into two groups with different colors based on whether the bicycle shop provides rental service or not. With the
processed location and other related information, we then
stored the files as readable XML files to be the source files
for our Toronto city data visualization.

TORONTO CITY MODEL AND
DATA SETS VISUALIZATION
With processed Toronto city model and data sets, we then
imported them into our HoloLens application. Because Holo76 IEEE Consumer Electronics Magazine

MAy 2018

Lens runs the Windows 10 OS, we used the Unity engine, C#,
and the library HoloToolKit for our implementation.
Because the source files of the Toronto city model are rather large, we scaled them down before importing to achieve a
better performance. The complete city model is made of several tiles, so we also selected a few tiles to build a subregion
of Toronto. In addition, a gray plane was placed underneath
the 3-D city model to improve the visibility. Figure 3 shows
the visualization result of the Toronto 3-D city model and the
menu to load city data files. After we selected Load File List,
we were able to view a list of city data types for visualization.
As previously mentioned, all open data sets were stored in
readable XML files with MTM coordinate and other related
information. We designed a menu above the city model to
choose the data set for visualization. Once a data set is selected, its data can be easily mapped on the corresponding locations on the 3-D city model. We also attached them together to
enable the data positions to change with scale adjustment of
the city model. Then cuboids with various heights and colors
were placed to represent different data properties. Figure 1 is
an example of visualizing Toronto city data on the 3-D city
model. All stations are represented with cuboids, whose
heights depend on the chosen value type. Red ones mean high
values, whereas blue ones are ordinary values. We can also
select other data sets from the menu to visualize, such as places of interest, cab stations, and bicycle shops.

INTERACTIONS DESIGN
Interactions between users and augmented contents is crucial
for the AR experience [17]. Thus, we integrated several
interaction designs into our implementation. HoloToolKit,
the official library provided by Microsoft, has already defined
several interaction types, such as gaze, gesture, and voice. We
used them to facilitate our development.
Gaze input can be used to select an object on HoloLens. It
displays a cursor at the center of the vision scope to determine the direction of a ray to hit the selected object. The
option on a menu selected by gaze input will usually change
size or color as feedback. We can also use gaze input to select
a specific building on the Toronto city model.
When it comes to the gesture input, the gestures we have
applied to interact with the city model and data objects are
called air-tap, manipulation, and navigation. Because gesture
input is detected by the camera embedded in the front part of
HoloLens, the user needs to put his/her hand in front of HoloLens to interact. The real finger in Figure 4 shows a user
using gesture to interact. When an option is selected, an airtap gesture can be applied to confirm. When we gaze at a
building on the Toronto city model, e.g., we can use an airtap to click it, and then the building will change to orange
and display detailed information with green sentences as the
feedback, shown in Figure 4. Manipulation and navigation
gestures can be used to rotate the city model, change its positions, adjust its size, and so on.
Voice commands were also used in our implementation,
and they can replace certain gesture inputs. The user can

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - May 2018