| 1 |
<h2>DESCRIPTION</h2> |
| 2 |
|
| 3 |
<em>i.rectify</em> uses the control points included in the source data |
| 4 |
or identified with the |
| 5 |
<a href="wxGUI.gcp.html">Ground Control Points Manager</a> |
| 6 |
to calculate a transformation matrix and then converts x,y |
| 7 |
cell coordinates to standard map coordinates for each pixel |
| 8 |
in the image. The result is a planimetric image with a |
| 9 |
transformed coordinate system (i.e., a different coordinate |
| 10 |
system than before it was rectified). Supported transformation methods |
| 11 |
are first, second, and third order polynomial and thin plate spline. |
| 12 |
Thin plate spline is recommended for ungeoreferenced satellite imagery |
| 13 |
where ground control points (GCPs) are included. Examples are |
| 14 |
<a href="https://gdal.org/en/stable/drivers/raster/l1b.html">NOAA/AVHRR</a> |
| 15 |
and <a href="https://gdal.org/en/stable/drivers/raster/esat.html#raster-esat">ENVISAT</a> |
| 16 |
imagery which include throusands of GCPs. |
| 17 |
|
| 18 |
<p> |
| 19 |
If no ground control points are available, the |
| 20 |
<a href="wxGUI.gcp.html">Ground Control Points Manager</a> |
| 21 |
must be run before <em>i.rectify</em>. An image must be |
| 22 |
georeferenced before it can reside in a standard coordinate |
| 23 |
project, and therefore be analyzed with the other map |
| 24 |
layers there. Upon |
| 25 |
completion of <em>i.rectify</em>, the rectified image is |
| 26 |
deposited in the target standard coordinate project. This |
| 27 |
project is selected using |
| 28 |
|
| 29 |
<em><a href="i.target.html">i.target</a></em>. |
| 30 |
|
| 31 |
<p>More than one raster map may be rectified at a time. Each cell file |
| 32 |
should be given a unique output file name. The rectified image or |
| 33 |
rectified raster maps will be located in the target project when the |
| 34 |
program is completed. The original unrectified files are not modified |
| 35 |
or removed. |
| 36 |
|
| 37 |
<p>If the <b>-c</b> flag is used, <em>i.rectify</em> will only rectify that |
| 38 |
portion of the image or raster map that occurs within the chosen window |
| 39 |
region in the target project, and only that portion of the cell |
| 40 |
file will be relocated in the target database. It is |
| 41 |
important therefore, to check the current mapset window in |
| 42 |
the target project if the <b>-c</b> flag is used. |
| 43 |
|
| 44 |
<p> |
| 45 |
If you are rectifying a file with plans to patch it to |
| 46 |
another file using the GRASS program <em>r.patch</em>, |
| 47 |
choose option number one, the current window in the target |
| 48 |
project. This window, however, must be the default window |
| 49 |
for the target project. When a file being rectified is |
| 50 |
smaller than the default window in which it is being |
| 51 |
rectified, NULLs are added to the rectified file. Patching |
| 52 |
files of the same size that contain NULL data, |
| 53 |
eliminates the possibility of a no-data line in the patched |
| 54 |
result. This is because, when the images are patched, the |
| 55 |
NULLs in the image are "covered" with non-NULL pixel |
| 56 |
values. When rectifying files that are going to be |
| 57 |
patched, rectify all of the files using the same default |
| 58 |
window. |
| 59 |
|
| 60 |
<h3>Coordinate transformation</h3> |
| 61 |
<p>The desired order of transformation (1, 2, or 3) is selected with the |
| 62 |
<b>order</b> option. |
| 63 |
|
| 64 |
The program will calculate the RMSE and check the required number of points. |
| 65 |
|
| 66 |
<h4>Linear affine transformation (1st order transformation)</h4> |
| 67 |
|
| 68 |
<dl> |
| 69 |
<dd> x' = ax + by + c |
| 70 |
<dd> y' = Ax + By + C |
| 71 |
</dl> |
| 72 |
|
| 73 |
The a, b, c, A, B, C are determined by least squares regression |
| 74 |
based on the control points entered. This transformation |
| 75 |
applies scaling, translation and rotation. It is NOT a |
| 76 |
general purpose rubber-sheeting like TPS, nor is it ortho-photo |
| 77 |
rectification using a DEM, not second order polynomial, |
| 78 |
etc. It can be used if (1) you have geometrically correct |
| 79 |
images, and (2) the terrain or camera distortion effect can |
| 80 |
be ignored. |
| 81 |
|
| 82 |
<h4>Polynomial Transformation Matrix (2nd, 3d order transformation)</h4> |
| 83 |
|
| 84 |
<em>i.rectify</em> uses a first, second, or third order transformation |
| 85 |
matrix to calculate the registration coefficients. The number |
| 86 |
of control points required for a selected order of transformation |
| 87 |
(represented by n) is |
| 88 |
|
| 89 |
<dl> |
| 90 |
<dd>((n + 1) * (n + 2) / 2) |
| 91 |
</dl> |
| 92 |
|
| 93 |
or 3, 6, and 10 respectively. It is strongly recommended |
| 94 |
that one or more additional points be identified to allow |
| 95 |
for an overly-determined transformation calculation which |
| 96 |
will generate the Root Mean Square (RMS) error values for |
| 97 |
each included point. The RMS error values for all the |
| 98 |
included control points are immediately recalculated when |
| 99 |
the user selects a different transformation order from the |
| 100 |
menu bar. The polynomial equations are performed using a |
| 101 |
modified Gaussian elimination method. |
| 102 |
|
| 103 |
<h4>Thin plate spline (TPS) transformation</h4> |
| 104 |
TPS transformation is selected with the <b>-t</b> flag. This method of |
| 105 |
coordinate transformation is recommended for satellite imagery where |
| 106 |
hundreds or thousands of GCPs are included, and for historical printed |
| 107 |
or scanned maps with unknown georeferencing and/or known localized |
| 108 |
distortions. |
| 109 |
<p> |
| 110 |
TPS combines a linear affine transformation with individual |
| 111 |
transformation coefficients for each GCP, using the radial basis kernel |
| 112 |
function with the distance <em>dist</em> between any two points: |
| 113 |
|
| 114 |
<dl> |
| 115 |
<dd>dist<sup>2</sup> * log(dist) |
| 116 |
</dl> |
| 117 |
|
| 118 |
As a consequence, localized distortions can be removed with TPS |
| 119 |
transformation. For example, scan line sensors will have due to the |
| 120 |
changing viewing angle larger distortions towards the end points of the |
| 121 |
scan line than at the center of the scan line. Even higher order |
| 122 |
polynomial transformations are not able to remove these locally |
| 123 |
different distortions, but TPS transformation can. For best results, |
| 124 |
TPS requires an even and, for localized distortions, dense spacing of |
| 125 |
GCPs. |
| 126 |
|
| 127 |
<h3>Resampling method</h3> |
| 128 |
<p>The rectified data is resampled with one of seven different methods: |
| 129 |
<em>nearest</em>, <em>bilinear</em>, <em>cubic</em>, <em>lanczos</em>, |
| 130 |
<em>bilinear_f</em>, <em>cubic_f</em>, or <em>lanczos_f</em>. |
| 131 |
<p>The <em>method=nearest</em> method, which performs a nearest neighbor assignment, |
| 132 |
is the fastest of the resampling methods. It is primarily used for |
| 133 |
categorical data such as a land use classification, since it will not change |
| 134 |
the values of the data cells. The <em>method=bilinear</em> method determines the new |
| 135 |
value of the cell based on a weighted distance average of the 4 surrounding |
| 136 |
cells in the input map. The <em>method=cubic</em> method determines the new value of |
| 137 |
the cell based on a weighted distance average of the 16 surrounding cells in |
| 138 |
the input map. The <em>method=lanczos</em> method determines the new value of |
| 139 |
the cell based on a weighted distance average of the 25 surrounding cells in |
| 140 |
the input map. |
| 141 |
<p>The bilinear, cubic and lanczos interpolation methods are most appropriate for |
| 142 |
continuous data and cause some smoothing. These options should not be used |
| 143 |
with categorical data, since the cell values will be altered. |
| 144 |
<p>In the bilinear, cubic and lanczos methods, if any of the surrounding cells used to |
| 145 |
interpolate the new cell value are NULL, the resulting cell will be NULL, even if |
| 146 |
the nearest cell is not NULL. This will cause some thinning along NULL borders, |
| 147 |
such as the coasts of land areas in a DEM. The <em>bilinear_f</em>, <em>cubic_f</em> |
| 148 |
and <em>lanczos_f</em> interpolation methods can be used if thinning along NULL edges is |
| 149 |
not desired. These methods "fall back" to simpler interpolation methods along NULL borders. |
| 150 |
That is, from lanczos to cubic to bilinear to nearest. |
| 151 |
<p>If nearest neighbor assignment is used, the output map has the same raster |
| 152 |
format as the input map. If any of the other interpolations is used, the |
| 153 |
output map is written as floating point. |
| 154 |
|
| 155 |
<p><!-- |
| 156 |
Note: In interactive mode it is possible to define a new file name |
| 157 |
for the target images. This is (currently) not provided in command line |
| 158 |
mode. |
| 159 |
--> |
| 160 |
|
| 161 |
<h2>NOTES</h2> |
| 162 |
|
| 163 |
If <em>i.rectify</em> starts normally but after some time the following text is seen: |
| 164 |
<br><code> |
| 165 |
ERROR: Error writing segment file |
| 166 |
</code><br> |
| 167 |
the user may try the <b>-c</b> flag or the module needs more free space |
| 168 |
on the hard drive. |
| 169 |
|
| 170 |
<h2>SEE ALSO</h2> |
| 171 |
|
| 172 |
The GRASS 4 <em> |
| 173 |
<a href="https://grass.osgeo.org/gdp/imagery/grass4_image_processing.pdf">Image |
| 174 |
Processing manual</a></em> |
| 175 |
|
| 176 |
<p><em> |
| 177 |
<a href="m.transform.html">m.transform</a>, |
| 178 |
<a href="r.proj.html">r.proj</a>, |
| 179 |
<a href="v.proj.html">v.proj</a>, |
| 180 |
<a href="i.group.html">i.group</a>, |
| 181 |
<a href="i.target.html">i.target</a> |
| 182 |
</em> |
| 183 |
<br> |
| 184 |
<em> |
| 185 |
<a href="wxGUI.gcp.html">Ground Control Points Manager</a> |
| 186 |
</em> |
| 187 |
|
| 188 |
<h2>AUTHORS</h2> |
| 189 |
|
| 190 |
William R. Enslin, |
| 191 |
Michigan State University, |
| 192 |
Center for Remote Sensing |
| 193 |
|
| 194 |
<p>Modified for GRASS 5.0 by:<br> |
| 195 |
Luca Palmeri ([email protected])<br> |
| 196 |
Bill Hughes<br> |
| 197 |
Pierre de Mouveaux ([email protected]) |
| 198 |
<br> |
| 199 |
CMD mode by Bob Covill |