I have a data analysis problem that I can reduce to one similar to analyzing the trajectories in the images below. These images show the tracks of subatomic particles interacting in a bubble chamber.
It's pretty obvious that by eye, easily discernible patterns can be seen. I want very much to know more about how classification and segmentation can be done using neural networks for this type of image.
These images are binary. The trajectory is either at a point in the image or it isn't. As can be seen, trajectories cross over one another, Some data appears to be missing in otherwise smooth curves, at arbitrary points along those curves. (My data may be more sparse in this respect.)
A typical paper on bubble chamber analysis that I would find deals with the analysis of the physics after trajectories have been classified and segmented.
Can anyone identify some papers that address this or something similar in the context of neural networks? I am not able to find anything recent on automated methods at all, but my google fu may not be up to the challenge. (By the way, I am less interested in some of the parametric methods like Hough Transforms. I'd like to focus on the neural approach.)
(I posted this previous question which wasn't quite as specific as this one. I hope there is some available research in this area related to physics that might give me some insights that are more directly related to my problem.)