A non-recursive border finding algorithm for linear quadtree based images
Given a digital binary image represented in a linear quadtree, its border is to be found as another linear quadtree. A new algorithm is proposed that can generate the border in sorted order. A pattern is found for the order of the edge pixels of a node. The differences of the location codes for t...
| Main Author: | |
|---|---|
| Language: | English |
| Published: |
2009
|
| Online Access: | http://hdl.handle.net/2429/5545 |
| Summary: | Given a digital binary image represented in a linear quadtree, its border is
to be found as another linear quadtree. A new algorithm is proposed that
can generate the border in sorted order. A pattern is found for the order
of the edge pixels of a node. The differences of the location codes for the
edge pixels of a node with grouping factor g is stored in a lookup table.
The lookup table also provides the information about which edge a given
pixel is on. Through a probabilistic analysis, the new algorithm is able to
avoid recursion which reduces the number of neighbour finding operations
significantly. The algorithm performs better, when compared to the one by
Yang & Lin, in both the run time (40% improvement) and the number of
neighbour finding operations incurred (60% reduction). |
|---|