1i.smap(1) Grass User's Manual i.smap(1)
2
6 i.smap - Performs contextual image classification using sequential
7 maximum a posteriori (SMAP) estimation.
8
10 imagery
11
13 i.smap
14 i.smap help
15 i.smap [-mq] group=string subgroup=string signaturefile=string [block‐
16 size=integer] output=string [--overwrite]
17 Flags:
19 -m Use maximum likelihood estimation (instead of smap)
20 -q Run quietly
22 --overwrite
24 Parameters:
26 group=string
27 imagery group
28 subgroup=string
30 imagery subgroup
31 signaturefile=string
33 imagery signaturefile
34 blocksize=integer
36 size of submatrix to process at one time Default: 128
37 output=string
39 output raster map
40
42 The i.smap program is used to segment multispectral images using a
43 spectral class model known as a Gaussian mixture distribution. Since
44 Gaussian mixture distributions include conventional multivariate Gauss‐
45 ian distributions, this program may also be used to segment multispec‐
46 tral images based on simple spectral mean and covariance parameters.
47 i.smap has two modes of operation. The first mode is the sequential
49 maximum a posteriori (SMAP) mode [2]. The SMAP segmentation algorithm
50 attempts to improve segmentation accuracy by segmenting the image into
51 regions rather than segmenting each pixel separately (see NOTES).
52 The second mode is the more conventional maximum likelihood (ML) clas‐
54 sification which classifies each pixel separately, but requires some‐
55 what less computation. This mode is selected with the -m flag (see
56 below).
57
59 Flags:
60 -m Use maximum likelihood estimation (instead of smap). Normal
61 operation is to use SMAP estimation (see NOTES).
62 -q Run quietly, without printing messages about program progress.
64 Without this flag, messages will be printed (to stderr) as the
65 program progresses.
66 Parameters:
68 group=name
69 imagery group
70 The imagery group that defines the image to be classified.
71 subgroup=name
73 imagery subgroup
74 The subgroup within the group specified that specifies the sub‐
75 set of the band files that are to be used as image data to be
76 classified.
77 signaturefile=name
79 imagery signaturefile
80 The signature file that contains the spectral signatures (i.e.,
81 the statistics) for the classes to be identified in the image.
82 This signature file is produced by the program i.gensigset (see
83 NOTES).
84 blocksize=value
86 size of submatrix to process at one time
87 default: 128
88 This option specifies the size of the "window" to be used when
89 reading the image data.
90 This program was written to be nice about memory usage without influ‐
92 encing the resultant classification. This option allows the user to
93 control how much memory is used. More memory may mean faster (or
94 slower) operation depending on how much real memory your machine has
95 and how much virtual memory the program uses.
96 The size of the submatrix used in segmenting the image has a principle
98 function of controlling memory usage; however, it also can have a sub‐
99 tle effect on the quality of the segmentation in the smap mode. The
100 smoothing parameters for the smap segmentation are estimated separately
101 for each submatrix. Therefore, if the image has regions with qualita‐
102 tively different behavior, (e.g., natural woodlands and man-made agri‐
103 cultural fields) it may be useful to use a submatrix small enough so
104 that different smoothing parameters may be used for each distinctive
105 region of the image.
106 The submatrix size has no effect on the performance of the ML segmenta‐
108 tion method.
109 output=name
111 output raster map.
112 The name of a raster map that will contain the classification
113 results. This new raster map layer will contain categories that
114 can be related to landcover categories on the ground.
115
117 If none of the arguments are specified on the command line, i.smap will
118 interactively prompt for the names of the maps and files.
119
121 The SMAP algorithm exploits the fact that nearby pixels in an image are
122 likely to have the same class. It works by segmenting the image at
123 various scales or resolutions and using the course scale segmentations
124 to guide the finer scale segmentations. In addition to reducing the
125 number of misclassifications, the SMAP algorithm generally produces
126 segmentations with larger connected regions of a fixed class which may
127 be useful in some applications.
128 The amount of smoothing that is performed in the segmentation is depen‐
130 dent of the behavior of the data in the image. If the data suggests
131 that the nearby pixels often change class, then the algorithm will
132 adaptively reduce the amount of smoothing. This ensures that exces‐
133 sively large regions are not formed.
134 The module i.smap does not support MASKed or NULL cells. Therefore it
136 might be necessary to create a copy of the classification results using
137 e.g. r.mapcalc.
138 r.mapcalc MASKed_map=classification results
140
142 C. Bouman and M. Shapiro, "Multispectral Image Segmentation using a
143 Multiscale Image Model", Proc. of IEEE Int'l Conf. on Acoust., Speech
144 and Sig. Proc., pp. III-565 - III-568, San Francisco, California, March
145 23-26, 1992. C. Bouman and M. Shapiro 1994, "A Multiscale Random Field
146 Model for Bayesian Image Segmentation", IEEE Trans. on Image Process‐
147 ing., 3(2), 162-177" McCauley, J.D. and B.A. Engel 1995, "Comparison of
148 Scene Segmentations: SMAP, ECHO and Maximum Likelyhood", IEEE Trans. on
149 Geoscience and Remote Sensing, 33(6): 1313-1316.
150
152 i.group for creating groups and subgroups
153 r.mapcalc to copy classification result in order to cut out MASKed sub‐
155 areas
156 i.gensigset to generate the signature file required by this program
158
160 Charles Bouman, School of Electrical Engineering, Purdue University
161 Michael Shapiro, U.S.Army Construction Engineering Research Laboratory
162 Last changed: $Date: 2004/09/01 15:35:50 $
164 Full index
166GRASS 6.2.2 i.smap(1)