Environmental Engineering Reference
In-Depth Information
As seen in Figure 3, the camera on the right is removed. Instead, a mirror is put on the edge
of the screen. For the convenience of illustration, we cut the width of the screen to half of
that is shown in Figure 2. The mirror is configured to be parallel to the left edge of the
screen, or the Y-axis of the screen-based coordinate system. Therefore we can have a mirror
image of the camera (virtual camera) with the exact same physical characters as the real one.
By combining the physical camera and its reflection (the virtual camera) together, we can
obtain a pseudo-stereovision.
From the set up shown in Figure 3, it is obvious that
2
1
(8)
2
1
The focal point or origin of the virtual camera frame is at [d
2x
, d
2y
]
T
=[2L
H
+2s
1x
, 0]
T
.
Meanwhile, as seen from Figure 3, the mirror image of the pointer P=[P
x
, P
y
]
T
is P
1=
[P
1x
,
P
1y
]
T
. Its projection to the image plane of the physical camera is N
2
. As depicted from Figure
3, the mirror image of the projection of P
1
on the physical camera is the projection of the
point P on the virtual camera. It is clear that N'
2
=-N
2
.
If we substitute the above values to Equation 5, we can easily find the position of pointer
P=[P
x
, P
y
]
T
and its corresponding cursor position P
s
=[P
sx
, P
sy
]
T
.
3.2 Monovision with mirror image
Alternatively, we can use a different approach. That is, instead of using the virtual camera
and convert back to the case of stereovision, we can process the mirror image of the pointer
at the physical camera directly.
Fig. 4. An alternative configuration of Pointer Localization Using Monovision with mirror
image.
