Use a Hough dual space of coordinate plots based on the parameters of helices (radius,slope,translational offset).

A similar problem for condensation droplets from gamma rays is solved very efficiently this way - the model for gamma rays is lines so you take the equation y=mx+b to plot condensations in m b coordinates (slope intercept parameter space) instead of x y coordinates. Next, you plot maxima vs radius of blurring, and apply Akaike equivalent model selection using weighting term to get the right number of solutions (a penalty for picking too many helices countered by a penalty for blurring too much).

I don’t know about state-of-the-art but would try to represent the particle track using angular and radial states, about an origin that is a polynomial in Cartesian space.

Hello,

Have you tried simple least square with a parametric analytic model ?

It could be perhaps work ?

But be careful when implementing least square : you should have to use SVD for good conditioning of the system.

Hope it may help you.

Hervé

Dear Pr. Kothapalli,

I think I have forgot a point in your problem.

In fact you should have a huge database of point and you want to build

the different helices corresponding to each charged particles ?

Is that correct ?

The generalized Hough transform as proposed Pr. Pearlman could be tried, but am not sure that for this dimension and the complexity of the model it will give accurate results ? I do not know.

I have made several experiments with those technics and it was not always relevant.

I think that other approaches could be investigate :

- Step 1 : find a clustering technic to divide the set of point in 3D in several subset where each one corresponds to 1 particle

(I think here a route for study is to detect outliers)

- Step 2 : least square as I have explaimed above.

I am very interested in your problem.

Could you provide your dataset ?

Cheers,

HP