X hits on this document

584 views

0 shares

56 / 170

# Black Spot Firmware

ROI1

ROI2

ROI1

ROI2

(a)

(b)

Figure 4.7 With a fixed ROI size, ROIs can overlap when the camera moves away from a beacon. This is because the number of pixels/m decreases as the distance from the camera increases. (a) shows two ROIs that do not overlap. (b) shows the same ROIs overlapping after the camera has moved away from the markers.

lgorithm 1 Pseudo code for a proposed algorithm to update ROI size is shown. FOR E CH FR ME FOR E CH ROI LET S REQUIRED = 2 * Vmax + D IF S CTU L < S REQUIRED THEN S CTU L = S REQUIRED + T/2 ENDIF IF S CTU L > S REQUIRED + T THEN S CTU L = S REQUIRED + T/2 ENDIF END LOOP END LOOP

ROI size, S REQUIRED is the required ROI size given by Equation 3.23. The tolerance T stops the ROI size changing rapidly. The algorithm ensures that the ROI remains between

the size given by Equation 3.23 and this size plus the tolerance value. of perhaps 5 pixels is suggested.

small value for T

t the time of writing the algorithm for automatically adjusting the ROI size has not been implemented.

4.2.4

# Marker position prediction

method to improve the relationship between the ROI size required for a given tracking speed involves predicting the position of the marker. If the position of the marker in the next frame can be estimated, larger marker speeds can be tracked with smaller ROI sizes. Consider a marker moving at a constant velocity across the camera’s image plane measured in pixels/frame. Since this velocity is constant, the position of the marker in the next frame can be predicted based on this velocity and using simple kinematic equations.

linear prediction algorithm was devised to track the marker positions. Using the marker

position at frame n and frame n

1, the average velocity of the marker between the two

 Document views 584 Page views 584 Page last viewed Sun Jan 22 19:00:20 UTC 2017 Pages 170 Paragraphs 6307 Words 54996