Class STRtree

  • All Implemented Interfaces:
    java.io.Serializable, SpatialIndex

    public class STRtree
    extends AbstractSTRtree
    implements SpatialIndex, java.io.Serializable
    A query-only R-tree created using the Sort-Tile-Recursive (STR) algorithm. For two-dimensional spatial data.

    The STR packed R-tree is simple to implement and maximizes space utilization; that is, as many leaves as possible are filled to capacity. Overlap between nodes is far less than in a basic R-tree. However, once the tree has been built (explicitly or on the first call to #query), items may not be added or removed.

    Described in: P. Rigaux, Michel Scholl and Agnes Voisard. Spatial Databases With Application To GIS. Morgan Kaufmann, San Francisco, 2002.

    Note that inserting items into a tree is not thread-safe. Inserting performed on more than one thread must be synchronized externally.

    Querying a tree is thread-safe. The building phase is done synchronously, and querying is stateless.

    Version:
    1.7
    See Also:
    Serialized Form
    • Constructor Summary

      Constructors 
      Constructor Description
      STRtree()
      Constructs an STRtree with the default node capacity.
      STRtree​(int nodeCapacity)
      Constructs an STRtree with the given maximum number of child nodes that a node may have.
    • Constructor Detail

      • STRtree

        public STRtree()
        Constructs an STRtree with the default node capacity.
      • STRtree

        public STRtree​(int nodeCapacity)
        Constructs an STRtree with the given maximum number of child nodes that a node may have.

        The minimum recommended capacity setting is 4.

    • Method Detail

      • insert

        public void insert​(Envelope itemEnv,
                           java.lang.Object item)
        Inserts an item having the given bounds into the tree.
        Specified by:
        insert in interface SpatialIndex
      • query

        public java.util.List query​(Envelope searchEnv)
        Returns items whose bounds intersect the given envelope.
        Specified by:
        query in interface SpatialIndex
        Parameters:
        searchEnv - the envelope to query for
        Returns:
        a list of the items found by the query
      • query

        public void query​(Envelope searchEnv,
                          ItemVisitor visitor)
        Returns items whose bounds intersect the given envelope.
        Specified by:
        query in interface SpatialIndex
        Parameters:
        searchEnv - the envelope to query for
        visitor - a visitor object to apply to the items found
      • remove

        public boolean remove​(Envelope itemEnv,
                              java.lang.Object item)
        Removes a single item from the tree.
        Specified by:
        remove in interface SpatialIndex
        Parameters:
        itemEnv - the Envelope of the item to remove
        item - the item to remove
        Returns:
        true if the item was found
      • size

        public int size()
        Returns the number of items in the tree.
        Returns:
        the number of items in the tree
      • depth

        public int depth()
        Returns the number of items in the tree.
        Returns:
        the number of items in the tree
      • nearestNeighbour

        public java.lang.Object[] nearestNeighbour​(ItemDistance itemDist)
        Finds the two nearest items in the tree, using ItemDistance as the distance metric. A Branch-and-Bound tree traversal algorithm is used to provide an efficient search.

        If the tree is empty, the return value is null If it is required to find only pairs of distinct items, the ItemDistance function must be anti-reflexive.

        Parameters:
        itemDist - a distance metric applicable to the items in this tree
        Returns:
        the pair of the nearest items or null if the tree is empty
      • nearestNeighbour

        public java.lang.Object nearestNeighbour​(Envelope env,
                                                 java.lang.Object item,
                                                 ItemDistance itemDist)
        Finds the item in this tree which is nearest to the given Object, using ItemDistance as the distance metric. A Branch-and-Bound tree traversal algorithm is used to provide an efficient search.

        The query object does not have to be contained in the tree, but it does have to be compatible with the itemDist distance metric.

        Parameters:
        env - the envelope of the query item
        item - the item to find the nearest neighbour of
        itemDist - a distance metric applicable to the items in this tree and the query item
        Returns:
        the nearest item in this tree or null if the tree is empty
      • nearestNeighbour

        public java.lang.Object[] nearestNeighbour​(STRtree tree,
                                                   ItemDistance itemDist)
        Finds the two nearest items from this tree and another tree, using ItemDistance as the distance metric. A Branch-and-Bound tree traversal algorithm is used to provide an efficient search. The result value is a pair of items, the first from this tree and the second from the argument tree.
        Parameters:
        tree - another tree
        itemDist - a distance metric applicable to the items in the trees
        Returns:
        the pair of the nearest items, one from each tree or null if no pair of distinct items can be found
      • isWithinDistance

        public boolean isWithinDistance​(STRtree tree,
                                        ItemDistance itemDist,
                                        double maxDistance)
        Tests whether some two items from this tree and another tree lie within a given distance. ItemDistance is used as the distance metric. A Branch-and-Bound tree traversal algorithm is used to provide an efficient search.
        Parameters:
        tree - another tree
        itemDist - a distance metric applicable to the items in the trees
        maxDistance - the distance limit for the search
        Returns:
        true if there are items within the distance
      • nearestNeighbour

        public java.lang.Object[] nearestNeighbour​(Envelope env,
                                                   java.lang.Object item,
                                                   ItemDistance itemDist,
                                                   int k)
        Finds k items in this tree which are the top k nearest neighbors to the given item, using itemDist as the distance metric. A Branch-and-Bound tree traversal algorithm is used to provide an efficient search. This method implements the KNN algorithm described in the following paper:

        Roussopoulos, Nick, Stephen Kelley, and Frédéric Vincent. "Nearest neighbor queries." ACM sigmod record. Vol. 24. No. 2. ACM, 1995.

        The query item does not have to be contained in the tree, but it does have to be compatible with the itemDist distance metric.

        Parameters:
        env - the envelope of the query item
        item - the item to find the nearest neighbour of
        itemDist - a distance metric applicable to the items in this tree and the query item
        k - the K nearest items in kNearestNeighbour
        Returns:
        the K nearest items in this tree