i.segment (1)
NAME
i.segment - Identifies segments (objects) from imagery data.
KEYWORDS
imagery, segmentation, classification, object recognition
SYNOPSIS
i.segment
i.segment --help
i.segment [-dw] group=name output=name threshold=float [method=string] [similarity=string] [minsize=integer] [memory=integer] [iterations=integer] [seeds=name] [bounds=name] [goodness=name] [--overwrite] [--help] [--verbose] [--quiet] [--ui]
Flags:
-d
Use 8 neighbors (3x3 neighborhood) instead of the default 4 neighbors for each pixel
-w
Weighted input, do not perform the default scaling of input raster maps
--overwrite
Allow output files to overwrite existing files
--help
Print usage summary
--verbose
Verbose module output
--quiet
Quiet module output
--ui
Force launching GUI dialog
Parameters:
group=name [required]
Name of input imagery group
output=name [required]
Name for output raster map
threshold=float [required]
Difference threshold between 0 and 1
Threshold = 0 merges only identical segments; threshold = 1 merges all
method=string
Segmentation method
Options: region_growing
Default: region_growing
similarity=string
Similarity calculation method
Options: euclidean, manhattan
Default: euclidean
minsize=integer
Minimum number of cells in a segment
The final step will merge small segments with their best neighbor
Options: 1-100000
Default: 1
memory=integer
Memory in MB
Default: 300
iterations=integer
Maximum number of iterations
Default: 20
seeds=name
Name for input raster map with starting seeds
bounds=name
Name of input bounding/constraining raster map
Must be integer values, each area will be segmented independent of the others
goodness=name
Name for output goodness of fit estimate map
DESCRIPTION
Image segmentation or object recognition is the process of grouping similar pixels into unique segments, also referred to as objects. Boundary and region based algorithms are described in the literature, currently a region growing and merging algorithm is implemented. Each object found during the segmentation process is given a unique ID and is a collection of contiguous pixels meeting some criteria. Note the contrast with image classification where all pixels similar to each other are assigned to the same class and do not need to be contiguous. The image segmentation results can be useful on their own, or used as a preprocessing step for image classification. The segmentation preprocessing step can reduce noise and speed up the classification.
NOTES
Region Growing and Merging
This segmentation algorithm sequentially examines all current segments in the raster map. The similarity between the current segment and each of its neighbors is calculated according to the given distance formula. Segments will be merged if they meet a number of criteria, including:
1 The pair is mutually most similar to each other (the similarity distance will be smaller than to any other neighbor), and
2 The similarity must be lower than the input threshold. The process is repeated until no merges are made during a complete pass.
Similarity and Threshold
The similarity between segments and unmerged objects is used to determine which objects are merged. Smaller distance values indicate a closer match, with a similarity score of zero for identical pixels.
During normal processing, merges are only allowed when the similarity between two segments is lower than the given threshold value. During the final pass, however, if a minimum segment size of 2 or larger is given with the minsize parameter, segments with a smaller pixel count will be merged with their most similar neighbor even if the similarity is greater than the threshold.
The threshold must be larger than 0.0 and smaller than 1.0. A threshold of 0 would allow only identical valued pixels to be merged, while a threshold of 1 would allow everything to be merged. Initial empirical tests indicate threshold values of 0.01 to 0.05 are reasonable values to start. It is recommended to start with a low value, e.g. 0.01, and then perform hierarchical segmentation by using the output of the last run as seeds for the next run.
Calculation Formulas
Both Euclidean and Manhattan distances use the normal definition, considering each raster in the image group as a dimension. In future, the distance calculation will also take into account the shape characteristics of the segments. The normal distances are then multiplied by the input radiometric weight. Next an additional contribution is added: (1-radioweight) * {smoothness * smoothness weight + compactness * (1-smoothness weight)}, where compactness = Perimeter Length / sqrt( Area ) and smoothness = Perimeter Length / Bounding Box. The perimeter length is estimated as the number of pixel sides the segment has.
Seeds
The seeds map can be used to provide either seed pixels (random or selected points from which to start the segmentation process) or seed segments. If the seeds are the results of a previous segmentation with lower threshold, hierarchical segmentation can be performed. The different approaches are automatically detected by the program: any pixels that have identical seed values and are contiguous will be assigned a unique segment ID.
It is expected that the minsize will be set to 1 if a seed map is used, but the program will allow other values to be used. If both options are used, the final iteration that ignores the threshold will also ignore the seed map and force merges for all pixels (not just segments that have grown/merged from the seeds).
Maximum number of starting segments
For the region growing algorithm without starting seeds, each pixel is sequentially numbered. The current limit with CELL storage is 2 billion starting segment IDs. If the initial map has a larger number of non-null pixels, there are two workarounds:
1 Use starting seed pixels. (Maximum 2 billion pixels can be labeled with positive integers.)
2 Use starting seed segments. (By initial classification or other methods.)
Boundary Constraints
Boundary constraints limit the adjacency of pixels and segments. Each unique value present in the bounds raster are considered as a MASK. Thus no segments in the final segmentated map will cross a boundary, even if their spectral data is very similar.
Minimum Segment Size
To reduce the salt and pepper affect, a minsize greater than 1 will add one additional pass to the processing. During the final pass, the threshold is ignored for any segments smaller then the set size, thus forcing very small segments to merge with their most similar neighbor.
Goodness of Fit
The goodness of fit for each pixel is calculated as 1 - distance of the pixel to the object it belongs to. The distance is calculated with the selected similarity method. A value of 1 means identical values, perfect fit, and a value of 0 means maximum possible distance, worst possible fit.
EXAMPLES
Segmentation of RGB orthophoto
This example uses the ortho photograph included in the NC Sample Dataset. Set up an imagery group:
i.group group=ortho_group input=ortho_2001_t792_1m@PERMANENT
Set the region to a smaller test region (resolution taken from input ortho photograph).
g.region -p raster=ortho_2001_t792_1m n=220446 s=220075 e=639151 w=638592
Try out a low threshold and check the results.
i.segment group=ortho_group output=ortho_segs_l1 threshold=0.02
From a visual inspection, it seems this results in too many segments. Increasing the threshold, using the previous results as seeds, and setting a minimum size of 2:
i.segment group=ortho_group output=ortho_segs_l2 threshold=0.05 seeds=ortho_segs_l1 min=2
i.segment group=ortho_group output=ortho_segs_l3 threshold=0.1 seeds=ortho_segs_l2
i.segment group=ortho_group output=ortho_segs_l4 threshold=0.2 seeds=ortho_segs_l3
i.segment group=ortho_group output=ortho_segs_l5 threshold=0.3 seeds=ortho_segs_l4
The output ortho_segs_l4 with threshold=0.2 still has too many segments, but the output with threshold=0.3 has too few segments. A threshold value of 0.25 seems to be a good choice. There is also some noise in the image, lets next force all segments smaller than 10 pixels to be merged into their most similar neighbor (even if they are less similar than required by our threshold):
Set the region to match the entire map(s) in the group.
g.region -p raster=ortho_2001_t792_1m@PERMANENT
Run i.segment on the full map:
i.segment group=ortho_group output=ortho_segs_final threshold=0.25 min=10
Processing the entire ortho image with nearly 10 million pixels took about 450 times more then for the final run.
Segmentation of panchromatic channel
This example uses the panchromatic channel of the Landsat7 scene included in the North Carolina sample dataset:
# create group with single channel
i.group group=singleband input=lsat7_2002_80
# set computational region to Landsat7 PAN band
g.region raster=lsat7_2002_80 -p
# perform segmentation with minsize=5
i.segment group=singleband threshold=0.05 minsize=5 \
output=lsat7_2002_80_segmented_min5 goodness=lsat7_2002_80_goodness_min5
# perform segmentation with minsize=100
i.segment group=singleband threshold=0.05 minsize=100
output=lsat7_2002_80_segmented_min100 goodness=lsat7_2002_80_goodness_min100
Original panchromatic channel of the Landsat7 scene
Segmented panchromatic channel, minsize=5
Segmented panchromatic channel, minsize=100
TODO
Functionality
o Further testing of the shape characteristics (smoothness, compactness), if it looks good it should be added. (in progress)
o Malahanobis distance for the similarity calculation.
Use of Segmentation Results
o Improve the optional output from this module, or better yet, add a module for i.segment.metrics.
o Providing updates to i.maxlik to ensure the segmentation output can be used as input for the existing classification functionality.
o Integration/workflow for r.fuzzy (Addon).
Speed
o See create_isegs.c
REFERENCES
This project was first developed during GSoC 2012. Project documentation, Image Segmentation references, and other information is at the project wiki.
Information about classification in GRASS is at available on the wiki.
SEE ALSO
g.gui.iclass, i.group, i.maxlik, i.smap, r.kappa
AUTHORS
Eric Momsen - North Dakota State University
Markus Metz (GSoC Mentor)
Last changed: $Date: 2016-07-01 00:17:58 +0200 (Fri, 01 Jul 2016) $
SOURCE CODE
Available at: i.segment source code (history)
Main index | Imagery index | Topics index | Keywords index | Graphical index | Full index
2003-2017 GRASS Development Team, GRASS GIS 7.2.1 Reference Manual