scroll down for instructions and explanations

This applet implements a solution of the edge continuity problem using the LEGION artificial neural network.

Click on the grid to form a black and white image consisting of no more than 3 discontinuous curves or line segments. To select orientation for each cell, click on it, and then type in its orientation angle (in degrees from 0 to 180). A negative number or a non-numeric value will result in absence of orientation/stimulation. After you are done creating the image, press the Run button, and you will see the network's oscillators become active. Eventually neurons within continuous curves will synchronize with each other while desynchronizing from neurons that belong to other edges. The final result is not a stable configuration, but rather a set of waves of activity alternating between disconnected regions. Thus the image is segmented temporally, the behavior of each region alternating between active (excited) and inactive (inhibited) states.

The network design is based on two papers:

1) Wang D.L. (1995): Emergent synchrony in locally coupled neural oscillators. IEEE Transactions on Neural Networks, vol. 6, 941-948.

2) Wersing, H., Steil, J. J., Ritter, H. (2001). "A competitive-layer model for feature binding and sensory segmentation" Neural Computation, 357-388.