Class BundleDetector

java.lang.Object
sc.fiji.snt.util.BundleDetector

public class BundleDetector extends Object
Detects regions where two paths run parallel to each other for a sustained distance, complementary to CrossoverFinder (which catches brief perpendicular near-crossings). Common use case: identifying axons bundled along the same nerve, or a path that was inadvertently hijacked by signal from a neighboring neurite running side-by-side.

This is a thin wrapper over CrossoverFinder: it reuses the entire grid/proximity/run-extraction pipeline and only inverts the angle filter. Where CrossoverFinder (with thetaMinDeg > 0) keeps only high-angle approaches (X-shaped crossings), BundleDetector uses thetaMaxDeg > 0 to keep only low-angle approaches (||-shaped sustained proximity). Defaults are also tuned for sustained runs: higher minRunNodes, similar proximity.

The output type is CrossoverFinder.CrossoverEvent: Callers should interpret the events as "bundled run" regions rather than crossings; the medianAngleDeg field will be small (parallel), and the index-window span will be wide (sustained).

Author:
Tiago Ferreira
See Also: