1i.pansharpen(1) Grass User's Manual i.pansharpen(1)
2
6 i.pansharpen - Image fusion algorithms to sharpen multispectral with
7 high-res panchromatic channels
8
10 imagery, fusion, sharpen, Brovey, IHS, HIS, PCA
11
13 i.pansharpen
14 i.pansharpen --help
15 i.pansharpen [-slr] red=name green=name blue=name pan=name output=base‐
16 name method=string bitdepth=integer [--overwrite] [--help] [--ver‐
17 bose] [--quiet] [--ui]
18 Flags:
20 -s
21 Serial processing rather than parallel processing
22 -l
24 Rebalance blue channel for LANDSAT
25 -r
27 Rescale (stretch) the range of pixel values in each channel to the
28 entire 0-255 8-bit range for processing (see notes)
29 --overwrite
31 Allow output files to overwrite existing files
32 --help
34 Print usage summary
35 --verbose
37 Verbose module output
38 --quiet
40 Quiet module output
41 --ui
43 Force launching GUI dialog
44 Parameters:
46 red=name [required]
47 Name of raster map to be used for <red>
48 green=name [required]
50 Name of raster map to be used for <green>
51 blue=name [required]
53 Name of raster map to be used for <blue>
54 pan=name [required]
56 Name of raster map to be used for high resolution panchromatic
57 channel
58 output=basename [required]
60 Name for output basename raster map(s)
61 method=string [required]
63 Method for pan sharpening
64 Options: brovey, ihs, pca
65 Default: ihs
66 bitdepth=integer [required]
68 Bit depth of image (must be in range of 2-30)
69 Options: 2-32
70 Default: 8
71
73 i.pansharpen uses a high resolution panchromatic band from a multispec‐
74 tral image to sharpen 3 lower resolution bands. The 3 lower resolution
75 bands can then be combined into an RGB color image at a higher (more
76 detailed) resolution than is possible using the original 3 bands. For
77 example, Landsat ETM has low resolution spectral bands 1 (blue), 2
78 (green), 3 (red), 4 (near IR), 5 (mid-IR), and 7 (mid-IR) at 30m reso‐
79 lution, and a high resolution panchromatic band 8 at 15m resolution.
80 Pan sharpening allows bands 3-2-1 (or other combinations of 30m resolu‐
81 tion bands like 4-3-2 or 5-4-2) to be combined into a 15m resolution
82 color image.
83 i.pansharpen offers a choice of three different ’pan sharpening’ algo‐
84 rithms: IHS, Brovey, and PCA.
85 For IHS pan sharpening, the original 3 lower resolution bands, selected
86 as red, green and blue channels for creating an RGB composite image,
87 are transformed into IHS (intensity, hue, and saturation) color space.
88 The panchromatic band is then substituted for the intensity channel
89 (I), combined with the original hue (H) and saturation (S) channels,
90 and transformed back to RGB color space at the higher resolution of the
91 panchromatic band. The algorithm for this can be represented as: RGB ->
92 IHS -> [pan]HS -> RGB.
93 With a Brovey pan sharpening, each of the 3 lower resolution bands and
94 panchromatic band are combined using the following algorithm to calcu‐
95 late 3 new bands at the higher resolution (example for band 1):
96 band1
97 new band1 = ----------------------- * panband
98 band1 + band2 + band3
99 In PCA pan sharpening, a principal component analysis is performed on
100 the original 3 lower resolution bands to create 3 principal component
101 images (PC1, PC2, and PC3) and their associated eigenvectors (EV), such
102 that:
103 band1 band2 band3
104 PC1: EV1-1 EV1-2 EV1-3
105 PC2: EV2-1 EV2-2 EV2-3
106 PC3: EV3-1 EV3-2 EV3-3
107 and
108 PC1 = EV1-1 * band1 + EV1-2 * band2 + EV1-3 * band3 - mean(bands 1,2,3)
109 An inverse PCA is then performed, substituting the panchromatic band
110 for PC1. To do this, the eigenvectors matrix is inverted (in this case
111 transposed), the PC images are multiplied by the eigenvectors with the
112 panchromatic band substituted for PC1, and mean of each band is added
113 to each transformed image band using the following algorithm (example
114 for band 1):
115 band1 = pan * EV1-1 + PC2 * EV1-2 + PC3 * EV1-3 + mean(band1)
116 The assignment of the channels depends on the satellite. Examples of
117 satellite imagery with high resolution panchromatic bands, and lower
118 resolution spectral bands include Landsat 7 ETM, QuickBird, and SPOT.
119
121 The module works for 2-bit to 30-bit images. All images are rescaled to
122 8-bit for processing. By default, the entire possible range for the
123 selected bit depth is rescaled to 8-bit. For example, the range of
124 0-65535 for a 16-bit image is rescaled to 0-255). The ’r’ flag allows
125 the range of pixel values actually present in an image rescaled to a
126 full 8-bit range. For example, a 16 bit image might only have pixels
127 that range from 70 to 35000; this range of 70-35000 would be rescaled
128 to 0-255. This can give better visual distinction to features, espe‐
129 cially when the range of actual values in an image only occupies a rel‐
130 atively limited portion of the possible range.
131 i.pansharpen temporarily changes the computational region to the high
132 resolution of the panchromatic band during sharpening calculations,
133 then restores the previous region settings. The current region coordi‐
134 nates (and null values) are respected. The high resolution panchromatic
135 image is histogram matched to the band it is replaces prior to substi‐
136 tution (i.e., the intensity channel for IHS sharpening, the low res
137 band selected for each color channel with Brovey sharpening, and the
138 PC1 image for PCA sharpening).
139 By default, the command will attempt to employ parallel processing,
140 using up to 3 cores simultaneously. The -s flag will disable parallel
141 processing, but does use an optimized r.mapcalc expression to reduce
142 disk I/O.
143 The three pan-sharpened output channels may be combined with d.rgb or
144 r.composite. Colors may be optionally optimized with i.colors.enhance.
145 While the resulting color image will be at the higher resolution in all
146 cases, the 3 pan sharpening algorithms differ in terms of spectral
147 response.
148
150 Pan sharpening comparison example
151 Pan sharpening of a Landsat image from Boulder, Colorado, USA (LANDSAT
152 ETM+ [Landsat 7] spectral bands 5,4,2, and pan band 8):
153 # R, G, B composite at 30m
154 g.region raster=p034r032_7dt20010924_z13_20 -p
155 d.rgb b=p034r032_7dt20010924_z13_20 g=lp034r032_7dt20010924_z13_40
156 r=p034r032_7dt20010924_z13_50
157 # i.pansharpen with IHS algorithm
158 i.pansharpen red=p034r032_7dt20010924_z13_50 green=p034r032_7dt20010924_z13_40
159 blue=p034r032_7dt20010924_z13_20 pan=p034r032_7dp20010924_z13_80
160 output=ihs321 method=ihs
161 # ... likewise with method=brovey and method=pca
162 # display at 15m
163 g.region raster=ihs542_blue -p
164 d.rgb b=ihs542_blue g=ihs542_green r=ihs542_red
165 Results:
167  R, G, B composite of Landsat at 30m  R, G, B composite of Brovey sharpened image at 15m
169  R, G, B composite of IHS sharpened image at 15m  R, G, B composite of PCA sharpened image at 15m"
171 Pan sharpening of LANDSAT ETM+ (Landsat 7)
174 LANDSAT ETM+ (Landsat 7), North Carolina sample dataset:
175 # original at 28m
176 g.region raster=lsat7_2002_10 -p
177 d.mon wx0
178 d.rgb b=lsat7_2002_10 g=lsat7_2002_20 r=lsat7_2002_30
179 # i.pansharpen with IHS algorithm
180 i.pansharpen red=lsat7_2002_30@PERMANENT \
181 green=lsat7_2002_20 blue=lsat7_2002_10 \
182 pan=lsat7_2002_80 method=ihs \
183 output=lsat7_2002_ihs
184 # display at 14.25m
185 g.region raster=lsat7_2002_ihs_red -p
186 d.erase
187 d.rgb r=lsat7_2002_ihs_red g=lsat7_2002_ihs_green b=lsat7_2002_ihs_blue
188 # compare before/after (RGB support in "Advanced"):
189 g.gui.mapswipe
190 # optionally color balancing:
191 i.colors.enhance r=lsat7_2002_ihs_red g=lsat7_2002_ihs_green b=lsat7_2002_ihs_blue
192
194 i.his.rgb, i.rgb.his, i.pca, d.rgb, r.composite
195
197 · Original Brovey formula reference unknown, probably...
198 Roller, N.E.G. and Cox, S., (1980). Comparison of Landsat MSS
199 and merged MSS/RBV data for analysis of natural vegetation.
200 Proc. of the 14th International Symposium on Remote Sensing of
201 Environment, San Jose, Costa Rica, 23-30 April, pp. 1001-1007
202 · Amarsaikhan, D., Douglas, T. (2004). Data fusion and multi‐
204 source image classification. International Journal of Remote
205 Sensing, 25(17), 3529-3539.
206 · Behnia, P. (2005). Comparison between four methods for data
208 fusion of ETM+ multispectral and pan images. Geo-spatial Infor‐
209 mation Science, 8(2), 98-103.
210 · Du, Q., Younan, N. H., King, R., Shah, V. P. (2007). On the
212 Performance Evaluation of Pan-Sharpening Techniques. Geoscience
213 and Remote Sensing Letters, IEEE, 4(4), 518-522.
214 · Karathanassi, V., Kolokousis, P., Ioannidou, S. (2007). A com‐
216 parison study on fusion methods using evaluation indicators.
217 International Journal of Remote Sensing, 28(10), 2309-2341.
218 · Neteler, M, D. Grasso, I. Michelazzi, L. Miori, S. Merler, and
220 C. Furlanello (2005). An integrated toolbox for image regis‐
221 tration, fusion and classification. International Journal of
222 Geoinformatics, 1(1):51-61 (PDF)
223 · Pohl, C, and J.L van Genderen (1998). Multisensor image fusion
225 in remote sensing: concepts, methods and application. Int. J.
226 of Rem. Sens., 19, 823-854.
227
229 Michael Barton (Arizona State University, USA)
230 with contributions from Markus Neteler (ITC-irst, Italy); Glynn
231 Clements; Luca Delucchi (Fondazione E. Mach, Italy); Markus Metz; and
232 Hamish Bowman.
233
235 Available at: i.pansharpen source code (history)
236 Main index | Imagery index | Topics index | Keywords index | Graphical
238 index | Full index
239 © 2003-2019 GRASS Development Team, GRASS GIS 7.8.2 Reference Manual
241GRASS 7.8.2 i.pansharpen(1)