The lo-pass image of Figure 9 is thresholded to produce a mask (Fig 10). Isolated patches in the mask will usually disappear after an application of the lo-pass filter used earlier (Fig 11) followed by another thresholding (Fig 12). If the mask in Figure 12 retains any unwanted patches, the size of every connected group of pixels is found, and all but the largest patch are discarded. The finished mask is applied to the raw data shown in Figure 7. The result is seen in Figure 13, which is used to compute a centroid. If the algorithm works as expected, Figure 13 should closely resemble the quantized Airy disk of Figure 6.
|
|
|
|
|
Figure 14 displays the raw noisy spot of Figure 7 with the true centroid shown as a blue circle, and the calculated centroid shown as a red plus sign. Figure 15 shows Figure 14 clipped by the finished mask.
