X hits on this document

235 views

0 shares

123 / 170

0

.0040

0

.0042

0

.0036

0

.0025

0

.0087 / 0.0007

0

.1139 / 0.0012

3.0000

0.0131

0.0037

0.0143

2.5000

0.0104

0.0031

0.0119

2.0000

0.0111

0.0040

0.0113

1.5000

0.0092

0.0042

0.0096

1.0000

0.0137 / 0.0079

0.0197 / 0.0040

0.0116 / 0.0089

0.5000

0.0521 / 0.0170

0.0711 / 0.0134

0.0684 / 0.0114

109

# Distance (m)

Mean std. dev. Y

2σ value

Mean std. dev. X

2σ value

Table 8.1 The mean spread in centroid positions appears to decrease as the camera until a distance of 1 m. At 0.5 m the spread increases again. The values in the last two rows were calculated before outliers were removed. The values to the outliers removed.

module nears the to the left of the ’/’ the right have had

less than 80) and was removed from the data set and the table recalculated. The recalcu- lated values are shown following the ‘/’ in the table. One possible explanation for this low intensity spot is a reflection of the rail, as observed on a number of occasions while collect- ing data. This shows that there is a need for the system to be able to remove poor quality data.

The results have a tighter spread with these outliers removed, however, the mean standard deviation in X at 0.5 m is still large. Using the adjusted results it appears that the mean spread in marker centroids decreases as the module nears the camera but then begins to increase again somewhere between 1 m and 0.5 m. t 0.5 m there are only three markers visible (after removing the outlier). The first marker has a higher standard deviation than

the other two and appears to be lifting the mean standard deviation. The intensity of this marker at 0.5 m was investigated revealing that the marker was saturated meaning some of the marker pixels were at a value of 255. This could explain the lower performance.

# Interestingly, the standard deviation in the Y direction is often slightly greater than the

standard deviation in the X direction. This is consistent throughout the measurements.

s

yet it is not known what causes this and this is an area for future research. this difference is small the performance of the system will not be affected.

However, as

nother data set was collected in order to produce a more detailed trend.

The beacon

was translated towards the Black Spot in 0.06 m increments starting at a distance of ap- proximately 3 m and ending at approximately 0.5 m. plot was produced of the distance versus the standard deviation in the X and Y directions comprising 40 points as shown in Figure 8.9. In Figure 8.9 the mean standard deviations of all visible markers are plotted. Outliers have been removed based on the markers intensity values. s the beacon moved towards the Black Spot, some markers disappeared from view and therefore the mean was calculated with a varying number of markers. The last data point was discarded as it was

 Document views 235 Page views 235 Page last viewed Tue Oct 25 16:34:32 UTC 2016 Pages 170 Paragraphs 6307 Words 54996