Tree compression and optimization with applications
read more
Citations
Point cloud attribute compression with graph transform
Efficient memory representation of XML documents
Efficient suffix trees on secondary storage
Graph-based compression of dynamic 3D point cloud sequences
Graph-Based Compression of Dynamic 3D Point Cloud Sequences
References
Compilers: Principles, Techniques, and Tools
A method for the construction of minimum-redundancy codes
Arithmetic coding for data compression
The Quadtree and Related Hierarchical Data Structures
An effective way to represent quadtrees
Frequently Asked Questions (11)
Q2. How do the authors eliminate pointers in a linear quadtree?
In linear quadtrees pointers are eliminated by storing pixels (black only) by using an encoding which reflects the successive quadrant subdivisions.
Q3. What is the alternative to simple sequential storage?
A good alternative to simple sequential storage is to use pixel trees which try to divide the picture into uniform areas, where adjacent pixels have the same colour, and to hierarchically organize these areas [9,10,12,38].
Q4. What is the way to save space and allow efficient traversal in binary trees?
A straightforward method for saving space and allowing efficient traversal in binary trees is to "thread" the tree so that isomorphic subtrees are stored only once.
Q5. Why is the cost of a search in each of these structures dominated by the largest?
Due to the double exponential growth of the sizes, the cost of a search in each of these structures isdominated by that of the largest.
Q6. What is the heuristic for calculating the harmonic decay property?
When the resulting matrix fulfilling the harmonic decay property is compressed by the first-fit method, a structure with space requirement O(n log log n) is obtained.
Q7. What is the simplest method to encode the production label of a rule?
One of the simplest methods encodes the production label of a rule, whose left-hand side non-terminal occurs on the left-hand side of r rules by log2 r bits that will uniquely specify the production which has been used in the substitution of the non-terminal.
Q8. What is the common way to search for a string in a trie?
It can be (1) null, (2) a pointer to an auxiliary table containing the strings currently represented by the trie, or (3) a pointer to another node in the trie.
Q9. How can the authors make a tree traversal more efficient?
The authors omit the details concerning the use of the directories, however the results show that it is possible to construct a structure which uses 2n+ο(n) space and in which a tree traversal requires O(log n) bit-accesses, or O(1) time if O(log n) consecutive bits can be manipulated at unit cost.
Q10. What is the length of the paths from leaves to the root?
A binary tree is said to be complete if all the internal nodes have two children and all the paths from leaves to the root are of equal length.
Q11. What is the problem of minimizing the number of sets needed for binary search?
Another problem is to minimize the worstcase binary search time under the restriction that the number of sets does not exceed a given bound.