This preview shows page 1. Sign up to view the full content.
Unformatted text preview: te
segmentation (equation 5.1): And the maximum region homogeneity conditions (equations 5.31, 5.32): Three basic approaches to region growing exist: region merging, region splitting and
split-and-merge region growing. Faculty of Engineering Robotics Technology MECH 4041 22
B.Eng (Hons.) Mechatronics S. Venkannah Mechanical and Production Engineering Department Region merging starts with an over-segmented image in which regions satisfy equation
5.31. Regions are merged to satisfy condition (5.32) as long as equation (5.31) remains
Region splitting is the opposite of region merging. Region splitting begins with an
under-segmented image which does not satisfy condition (5.31). Therefore the existing
image regions are sequentially split to satisfy conditions (5.1), (5.31), and (5.32)
A combination of splitting and merging may result in a method with the advantages of
both other approaches. Split and merge approaches typically use pyramid image
representations. Because both split-and-merge processing options are available, the
starting segmentation does not have to satisfy either condition (5.31) or (5.32).
In watershed segmentation, catchment basins represent the regions of the segmented
image. The first, watershed segmentation approach starts with finding a downstream path
from each pixel of the image to the local minima of image surface altitude. A catchment
basin is then defined as the set of pixels for which their respective downstream paths all
end up in the same altitude minimum. In the second approach, each gray level minimum
represents one catchment basin and the strategy is to start filling the catchment basins
from the bottom.
Images segmented by region growing methods often contain either too many regions
(under growing) or too few regions (over-growing) as a result of non-optimal parameter
setting. To improve classification results, a variety of post-processors has been
developed. Simpler post-processors decrease the number of small regions in the
More complex post-processing may combine segmentation
information obtained from region growing and edge...
View Full Document
- Spring '14