Hardware: For the transmitter, three different screens are used to display the
trans-mitted code: a 24” flat display Dell P2416d、iPhone 6+ which has a 5.5” display and Apple watch series 4 with 1.8” display. The comparison of these three transmitters is shown in Table 6.1. For the receiver, to understand how the resolution of a camera and the quality of the image sensor affect the performance of the our proposed solutions, two cameras with different resolutions and image sensors are used in the experiments: a Point-Grey Flea3 (RL3-U3-88S2C-C) and a Canon 70D is used. The comparison of these two cameras is listed in Table 6.2. We first select a lens with 16 mm focal length for Flea 3. In order to maintain a similar field of view (FOV) between these two different type of cam-eras, we choose the lens whose the focal length is determined according to the following equation:
α = 2arctan d
2f, (6.1)
where α represents the angle of FOV, d represents the size of the image sensor and f is the focal length of the adopted lens. By substituting the image sensor size of Canon 70D into this equation, we can derive the focal length for Canon 70D, which is 50 mm.
Table 6.1: Transmitter specifications
DELL P2416d iPhone 6+ Apple watch series 4
Resolution 2560*1440 2560*1440 448*368
Pixel Pitch (mm) 0.205*0.205 0.063*0.063 Not found
Brightness (cd/m2) 300 566 Not found
Contrast Ratio 1000:1 1300:1 Not found
Table 6.2: Receiver specifications
Flea3 Canon 70D
Image Sensor Format Sony IMX121 (1 / 2.5”) Canon APS-C
Type of Sensor CMOS CMOS
Resolution 4096*2160 5472*3648
Pixel Size (µm) 1.55*1.55 4.1*4.1
Symbol size of the transmitted code: For Dell P2416D and iPhone 6 plus, the symbol size is set to be 101, corresponding to a data region of 242x242 pixels. For Apple watch, since its display is much smaller than the former ones, we use a symbol size of 41, referring to a 102x102 pixels data region. On the other hand, since the luminance level of the transmitter greatly influences the system performance, we set the luminance level of the transmitter to the maximum level. In addition, as the pixel size of the transmitter is too small, resulting in low luminance output, we combine 2x2 pixels as a macro pixel for displaying the transmitted code.
Performance metric: Most of the past works evaluate their performance with the use of BER at different communication distances. The main problem of this evaluation method, however, is that we cannot compare the performance of the same transmitted code on different transmitters, as the transmitted code on a screen with small pixel pitch is indeed smaller than that displayed on a screen with bigger pixel pitch. Since the size of displayed transmitted code has a significant influence on the size of the received image, at a same communication distance, the decoding performance with a smaller transmitted code would be worse than a bigger one. To enable a fair comparison across various transmitters,
in this thesis, we define a ratio of the size of received data region and the size of the transmitted data region as magnification ratio which can be expressed as:
magnif icationratio = Pr
Pt, (6.2)
where Pris the size of the received data region in pixel and Ptis the size of the transmitted data region in pixel. As the observed size of the displayed transmitted code decreases when the distance between the transmitter and the receiver increases, for the same transmitter, the smaller magnification ratio represents that distance between the transmitter and the receiver is longer.
Experimental setup: In the experiments, we setup a screen-to-camera link in an indoor and static scenario, as shown in Figure 6.2. Since the size of the transmitters are different, to maintain a similar magnification for comparison, the system performance of each transmitters are measured at different scales of distance.
1. Dell P2416D (Figure 6.1(b)): We measure the system performance using Dell monitor with a magnification ratio ranging from about 4 to 0.8, corresponding to a distance of 1.15∼4.4 m. The error rate is measured for every 20 cm.
2. iPhone 6+ (Figure 6.1(c)): We measure the system performance using iPhone with a magnification ratio ranging from about 3.8 to 1.1, corresponding to a distance of 0.43∼1.8 m. The error rate is measured for every 10 cm.
3. Apple watch (Figure 6.1(d)): We measure the system performance using Apple watch with a magnification ratio ranging from about 4 to 0.8, corresponding to a distance of 0.54∼1.6 m. The error rate is measured for every 10 cm.
At each distance, the transmitter is placed in the center of the FOV of the receiving camera. To enable the maximum system capability, we develop an algorithm to ensure the exposure setting of the camera is configured to have the best receiving performance (i.e., the maximum dynamic range) under the observed scene for each transmitter and receiver pair. To do so, we select several exposure settings and measure their responding curve of the intensity values between the screen and the camera with the obtained gradient patterns.
The dynamic range is defined as the difference between the maximum and the minimum
value of the responding curve. Among the measurement result, the setting with the largest dynamic range would be selected as our camera exposure setting. Three kinds of symbols modulated with BPSK, QPSK, and 8PSK are tested in our experiments.
(a) An illustration of the experimental environ-ment.
(b) Dell P2416D
(c) iPhone 6+ (d) Apple watch series 4
Figure 6.1: Experimental environment.