Detection Module

The tracker is automatically initialized on the first two frames by the detector. Worm detection generally involves:

• Binarization and filling of small holes: for images with a cluttered background, a background image is computed first (as shown in Figure 1; the worm in the center of the image is not moving, and so it is included in the background image), which will be subtracted from the original image to remove static spots.
• Worm boundary extraction based on the binary image
• Head and tail detection
• Center line and width computation
• Instantiation of the worm model.

In the detection module, detection is followed by the worm model validation to remove errors caused by spots in the background or overlapping worms. After the automatic initialization, the worm models will be instantiated. The head and tail also will be determined based on the direction of axial progression and amplitude of radial displacement. The detection is also used for correcting mis-tracked worms during the tracking process.

Figure 1: Background Image

Figure 2(a): Initialization by detection on an image with a clean background

Figure 2(b): Initialization by detection on an image with a cluttered background