Skip to main content
Figure 1 | EURASIP Journal on Image and Video Processing

Figure 1

From: A two-fly tracker that solves occlusions by dynamic programming: computational analysis of Drosophila courtship behaviour

Figure 1

Schematic overview of the image processing part. Preprocessor (red shape): (a) extract median background; (b) illumination correction curve; (c) chamber rejected in movement detection step; (d) accumulated peaks for circular arena detection; (e) rigorously smoothed background for body segmentation; (f) cautiously smoothed background for wing segmentation; (g) watched boundary for each detected arena; (h) chambers rejected by quality control, witness pictures capture reason for rejection; and (i) individual chambers after arena splitting step. Tracker (yellow shape): (j) gray values of an original frame of a single chamber video; (k) subtracting rigorously smoothed background for body region segmentation; (l) threshold determination: body region threshold (red line) and wing region threshold (green line) are determined within the gray value histogram (black line), respectively, a smoothed histogram (blue line); (m) binarized fly body region (n) gradient magnitude values, bright values indicate edges; (o) mask region for morphological reconstruction; (p) subtracting cautiously smoothed background for wing region segmentation, body and non-arena regions are black; (q) binarized wing region with legs; (r) fly wing region without legs and with filled holes; (s) original image with segmentation results, body region in red, wing region in blue; (t) cautiously smoothed background; and (u) rigorously smoothed background.

Back to article page