Computational Geometry Piyush Kumar (Lecture 5: Range Searching) Welcome to CIS5930 Range Searching : recap 1D Range search kD trees Range Trees Fractional Cascading Range Searching Preprocess a set P of objects for efficiently answering queries. Typically, P is a collection of geometric objects (points, rectangles, polygons) in Rd. Query Range, Q : d-rectangles, balls, halfspaces, simplices, etc.. Either count all objects in P Q or report the objects themselves. Courtesy Bhosle Example: Points in R2 Q1 Q2 Applications Databases Spatial databases (G.I.S.) Computer Graphics Robotics Vision Range Trees (1D) 6 2 4 5 17 7 8 12 15 7 19 4 Counting ? Reporting? 12 2 2 5 4 5 8 7 8 15 12 15 19 Range Trees (2d) P1 P2 P3 P4 P5 P6 P7 P8 Query and Space Complexity (counting) 1D Query : O(log n) Space: O(n) 2D Query : O(log n) Space O(nlogn) Construction time : O(nlogn) d-D Query : O(logd-1n) Space: O(nlogd-1n) Construction time : O(nlogd-1n) Kd-Trees a typical struct Int cut_dim; Double cut_val; Int size; // dim orthogonal to cutting plane // location of cutting plane // number of points in subtree The best implementation of kd-trees I know of: http://www.cs.umd.edu/~mount/ANN/ Look at kd_tree.cc and kd_tree.h kD-Trees (k-dimensional Trees) 1-d tree : split along median point and recursively build subtrees for the left and right sets. Higher dimensions : same approach, but cycle through the dimensions. Or, select the next dimension as the one with the widest spread. Efficiency of query processing drops as dimensions increase (becomes almost linear). However, the space requirement remains linear : O(n.d) kD-Trees (Contd.) c o m d f l n a b g e j k h i f k h a n b d o i c j g l e m kD-Trees (Contd.) Query complexity : How many cells can a query box intersect ? Let us consider a facet of the query • Any axis parallel line can intersect atmost 2 of these 4 cells. • Each of these 4 cells contain exactly n/4 points. Q(n) = 2.Q(n/4) + 1 Q(n) = O(n1/2) i.e. Query answered in O(n1-1/d + m) time where m is the output size Kd-Tree Summarizing Building (preprocessing time): O(n log n) Size: O(n) Range queries: O(sqrt(n)+k) In higher dimensions Building: O(dnlogn) Size: O(dn) Counting Query: O(n1-1/d) QuadTrees 4-way partitions Linear space Used in real life more than kd-trees Octrees 3D Version of Quadtrees 8 child nodes Applications Range searching Collision detection Mesh generation Visibility Culling Computer games General Sets of points We assumed till now that the x,y coordinates of the points are distinct and do not overlap the coordinates of the query. How do we relax this? More on the Chalk board…