10
Internet of Things (IoT)
defined for finding the intensity level in the image. For better accuracy, two or more
points are identified within the window region and the average intensity of the pixels is
used. Then, this intensity value is represented by their intensity percentage. According
to the database available for VTL percentage for different environmental conditions,
the approximate tinting level is determined. Table 1.1 shows the various tinting level on
green/blue channels and the corresponding RGB level of region in the image.
The second technique used is known as contour detection which improves the tint
detection of windscreen/window of vehicle. Contours can be explained simply as a
curve joining all the continuous points (along the boundary), having the same color
or intensity. In this technique, first we find the edges of the image by using relevant
edge detection technique. After edge detection, the image becomes a function of two
variables which are curves joining all continuous points, and these curves are called
contours. These contours are in different numbers depending on the reflection of light
through windshield/window glass. The number of contours is counted in the given
image. From the available database (calculated manually) of contour for various intensity levels, the threshold (here is the number of contours) is defined and is used for
determining the presence or absence of tint. If the number of contours is below the
specified threshold limit, then the tested windshield/window tinting is not allowed
according to the norms, and the process will switch to the next module, namely number
plate detection.
The third technique is called “Histogram analysis” and is used for visually judging
whether the image is in an appropriate range of gray level. Ideally, digital image should
use all available gray-scale range, from minimum to maximum. From histogram of tint
image (final extracted image), we can judge the presence of tinting on windshield/ window.
If the histogram of the extracted image has only two peaks of gray level, then it has undesired tinting level. If the histogram of the extracted image shows multiple peaks in the
histogram, then we can conclude that there is desirable level of tinting. These three methods collectively can determine the tinting level of window/windshield region. Figure 1.8
shows the flow chart of various tint level detection algorithms used in our present work.
After identifying the tinting level of vehicle window/windshield using these three
techniques, the process is switched to number plate identification module. If tinting
level is more than the desired level (prescribed norms by state/country), then it switches
to number plate identification module; otherwise, the test ends for the current vehicle.
After identifying the license plate registration number of a given vehicle, the system
TABLE 1.1
Various Tinting Level on Green/Blue Channels and Corresponding
RGB% Level Present in the Image
Tint VTL%
Green(grass) Value
Blue(sky) Value
R%
G%
B%
R%
G%
B%
No tint
49
55
20
67
75
68
50%
30
34
12
57
60
67
35%
16
22
8
40
40
40
30%
14
14
6
25
25
29
20%
9
9
3
19
19
21
15%
6
6
2
12
12
13
5%
0.4
0.4
0.4
0.4
0.4
0.4
Internet of Things (IoT)
defined for finding the intensity level in the image. For better accuracy, two or more
points are identified within the window region and the average intensity of the pixels is
used. Then, this intensity value is represented by their intensity percentage. According
to the database available for VTL percentage for different environmental conditions,
the approximate tinting level is determined. Table 1.1 shows the various tinting level on
green/blue channels and the corresponding RGB level of region in the image.
The second technique used is known as contour detection which improves the tint
detection of windscreen/window of vehicle. Contours can be explained simply as a
curve joining all the continuous points (along the boundary), having the same color
or intensity. In this technique, first we find the edges of the image by using relevant
edge detection technique. After edge detection, the image becomes a function of two
variables which are curves joining all continuous points, and these curves are called
contours. These contours are in different numbers depending on the reflection of light
through windshield/window glass. The number of contours is counted in the given
image. From the available database (calculated manually) of contour for various intensity levels, the threshold (here is the number of contours) is defined and is used for
determining the presence or absence of tint. If the number of contours is below the
specified threshold limit, then the tested windshield/window tinting is not allowed
according to the norms, and the process will switch to the next module, namely number
plate detection.
The third technique is called “Histogram analysis” and is used for visually judging
whether the image is in an appropriate range of gray level. Ideally, digital image should
use all available gray-scale range, from minimum to maximum. From histogram of tint
image (final extracted image), we can judge the presence of tinting on windshield/ window.
If the histogram of the extracted image has only two peaks of gray level, then it has undesired tinting level. If the histogram of the extracted image shows multiple peaks in the
histogram, then we can conclude that there is desirable level of tinting. These three methods collectively can determine the tinting level of window/windshield region. Figure 1.8
shows the flow chart of various tint level detection algorithms used in our present work.
After identifying the tinting level of vehicle window/windshield using these three
techniques, the process is switched to number plate identification module. If tinting
level is more than the desired level (prescribed norms by state/country), then it switches
to number plate identification module; otherwise, the test ends for the current vehicle.
After identifying the license plate registration number of a given vehicle, the system
TABLE 1.1
Various Tinting Level on Green/Blue Channels and Corresponding
RGB% Level Present in the Image
Tint VTL%
Green(grass) Value
Blue(sky) Value
R%
G%
B%
R%
G%
B%
No tint
49
55
20
67
75
68
50%
30
34
12
57
60
67
35%
16
22
8
40
40
40
30%
14
14
6
25
25
29
20%
9
9
3
19
19
21
15%
6
6
2
12
12
13
5%
0.4
0.4
0.4
0.4
0.4
0.4