KD tree construction

A k-d tree organizes points in space by repeatedly slicing the plane with straight cuts — first vertical, then horizontal, alternating at every level — always cutting at the median point so each side holds about half the points and the tree stays balanced. The real payoff is search: to find the point nearest to a query, you dive down to its region, remember the closest point as a "search circle," then backtrack and throw away any whole rectangle whose boundary is farther than that circle — because nothing inside it could possibly be closer. In this clip the tree examines only 5 of 13 points and prunes the other 8, which is exactly why this structure powers fast nearest-neighbor and range queries behind map search, image matching, game collision checks, and machine-learning classifiers.

#algorithm #kdtree #datastructures #nearestneighbor #computerscience #coding #spatial #search #visualization #programming

Видео KD tree construction канала Algo Vis