| 1 |
\ |
| 2 |
.\" This man page was generated by the Netpbm tool 'makeman' from HTML source. |
| 3 |
.\" Do not hand-hack it! If you have bug fixes or improvements, please find |
| 4 |
.\" the corresponding HTML page on the Netpbm website, generate a patch |
| 5 |
.\" against that, and send it to the Netpbm maintainer. |
| 6 |
.TH "pamhomography" 1 "04 December 2022" "netpbm documentation" |
| 7 |
|
| 8 |
|
| 9 |
|
| 10 |
|
| 11 |
<script type="text/javascript" src="https://polyfill.io/v3/polyfill.min.js?features=es6"></script> |
| 12 |
<script type="text/javascript" id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script> |
| 13 |
.SH NAME |
| 14 |
.PP |
| 15 |
pamhomography - stretch/shrink a quadrilateral region of an image to |
| 16 |
another quadrilateral shape |
| 17 |
|
| 18 |
|
| 19 |
.UN SYNOPSIS |
| 20 |
.SH SYNOPSIS |
| 21 |
.PP |
| 22 |
\fBpamhomography\fP |
| 23 |
[\fB-from\fP=\fIcoords\fP] |
| 24 |
[\fB-to\fP=\fIcoords\fP] |
| 25 |
[\fB-mapfile\fP=\fImap_file\fP] |
| 26 |
[\fB-view\fP=\fIcoords\fP] |
| 27 |
[\fB-fill\fP=\fIcolor\fP] |
| 28 |
[\fB-verbose\fP |
| 29 |
[\fIpam_file\fP] |
| 30 |
.PP |
| 31 |
You can abbreviate any option to its shortest unique prefix. You can use |
| 32 |
two hyphens instead of one to delimit an option. You can separate an option |
| 33 |
from its value with whitespace instead of \f(CW=\fP. |
| 34 |
|
| 35 |
|
| 36 |
.UN DESCRIPTION |
| 37 |
.SH DESCRIPTION |
| 38 |
.PP |
| 39 |
This program is part |
| 40 |
of |
| 41 |
.UR http://netpbm.sourceforge.net/ |
| 42 |
Netpbm |
| 43 |
.UE |
| 44 |
\&. |
| 45 |
.PP |
| 46 |
\fBpamhomography\fP stretches and shrinks an arbitrary quadrilateral |
| 47 |
portion of an input image you specify (not necessarily rectangular), into a |
| 48 |
new quadrilateral shape you specify, producing a new image. |
| 49 |
.PP |
| 50 |
You can do any |
| 51 |
.UR https://en.wikipedia.org/wiki/Affine_transformation#Image_transformation |
| 52 |
affine image transformation |
| 53 |
.UE |
| 54 |
\&: translation, reflection, scaling, |
| 55 |
rotation, and shearing/skewing. However, \fBpamhomography\fP additionally can |
| 56 |
do \fIbilinear\fP transforms, which means it can warp any quadrilateral to any |
| 57 |
other quadrilateral, even when this mapping cannot be described using a single |
| 58 |
set of linear equations. This can be useful, for example, for creating |
| 59 |
perspective views of rectangular images or for reverse-mapping a perspective |
| 60 |
view back to a rectangular projection. |
| 61 |
|
| 62 |
|
| 63 |
.UN OPTIONS |
| 64 |
.SH OPTIONS |
| 65 |
.PP |
| 66 |
In addition to the options common to all programs based on libnetpbm (most |
| 67 |
notably \fB-quiet\fP, see |
| 68 |
.UR index.html#commonoptions |
| 69 |
Common Options |
| 70 |
.UE |
| 71 |
\&), \fBpamhomography\fP recognizes the following command line |
| 72 |
options: |
| 73 |
|
| 74 |
|
| 75 |
|
| 76 |
.TP |
| 77 |
\fB-from\fP=\fIcoords\fP |
| 78 |
.sp |
| 79 |
This defines the source quadrilateral. \fIcoords\fP is a list of four |
| 80 |
integer-valued (\fIx\fP, \fIy\fP) coordinates. If you do not specify the |
| 81 |
source with either \fB-from\fP or \fB-mapfile\fP, the source quadrilateral |
| 82 |
is the entire input image. |
| 83 |
|
| 84 |
|
| 85 |
.TP |
| 86 |
\fB-to\fP=\fIcoords\fP |
| 87 |
.sp |
| 88 |
This defines the target quadrilateral. \fIcoords\fP is a list of four |
| 89 |
integer-valued (\fIx\fP, \fIy\fP) coordinates. If you do not specify the |
| 90 |
target with either \fB-to\fP or \fB-mapfile\fP, the target quadrilateral is |
| 91 |
the same as the entire input image. |
| 92 |
|
| 93 |
.TP |
| 94 |
\fB-mapfile\fP=\fImap_file\fP |
| 95 |
.sp |
| 96 |
This names a text file that describes the mapping from the source to the |
| 97 |
target quadrilateral. The file \fImap_file\fP must contain either eight |
| 98 |
integer-valued (\fIx\fP, \fIy\fP) coordinates, being the four source |
| 99 |
coordinates followed by the corresponding four target coordinates, or only |
| 100 |
four (\fIx\fP, \fIy\fP) coordinates, being only the four target |
| 101 |
coordinates. In the latter case, the source quadrilateral is taken to be the |
| 102 |
four corners of the input image in clockwise order, starting from the upper |
| 103 |
left. |
| 104 |
.sp |
| 105 |
Anything you specify with \fB-to\fP or \fB-from\fP overrides what is in |
| 106 |
this file. |
| 107 |
|
| 108 |
|
| 109 |
.TP |
| 110 |
\fB-view\fP=\fIcoords\fP |
| 111 |
.sp |
| 112 |
This defines the target view. \fIcoords\fP is a list of two integer-valued |
| 113 |
(\fIx\fP, \fIy\fP) coordinates: the upper left and lower right boundaries, |
| 114 |
respectively, of the pixels that will be visible in the output image. If |
| 115 |
\fB-view\fP is not specified, the target view will fit precisely the target |
| 116 |
quadrilateral. |
| 117 |
|
| 118 |
|
| 119 |
.TP |
| 120 |
\fB-fill\fP=\fIcolor\fP |
| 121 |
.sp |
| 122 |
This is the color with which the program fills all pixels that lie outside |
| 123 |
of the target quadrilateral. Specify the color as described for the |
| 124 |
.UR libnetpbm_image.html#colorname |
| 125 |
argument of the pnm_parsecolor() library routine |
| 126 |
.UE |
| 127 |
\&. |
| 128 |
.sp |
| 129 |
The default is black, and for images with a transparency plane, transparent. |
| 130 |
|
| 131 |
|
| 132 |
|
| 133 |
.TP |
| 134 |
\fB-verbose\fP |
| 135 |
This makes the program issue some informational messages about what it is |
| 136 |
doing. |
| 137 |
|
| 138 |
|
| 139 |
.PP |
| 140 |
Cooordinates should normally be specified in clockwise order. The syntax is |
| 141 |
fairly flexible: all characters other than the plus sign, minus sign, and |
| 142 |
digits are treated as separators. Although coordinates need to be integers, |
| 143 |
they may lie outside the image's boundary. |
| 144 |
|
| 145 |
.UN PARAMETERS |
| 146 |
.SH PARAMETERS |
| 147 |
.PP |
| 148 |
\fBpamhomography\fP's only parameter, \fIpam_file\fP, is the name of the |
| 149 |
file containing the input image. If you don't specify \fIpam_file\fP, the |
| 150 |
image comes from Standard Input. |
| 151 |
|
| 152 |
|
| 153 |
.UN NOTES |
| 154 |
.SH NOTES |
| 155 |
.PP |
| 156 |
The output image uses the same Netpbm format as the input image. |
| 157 |
.PP |
| 158 |
Simple transformations are best handled by other Netpbm programs, such as |
| 159 |
those listed in the |
| 160 |
.UR #SEE-ALSO |
| 161 |
\&'SEE ALSO' |
| 162 |
.UE |
| 163 |
\& section |
| 164 |
below. Use \fBpamhomography\fP for more sophisticated transformations such as |
| 165 |
perspective adjustments, rotations around an arbitrary point in the image, |
| 166 |
extraction of non-rectangular quadrilaterals, shearings by coordinates rather |
| 167 |
than by angle, and, in general, all transformations that are most easily |
| 168 |
expressed as mapping four points in one image to four points in another |
| 169 |
image. |
| 170 |
|
| 171 |
.UN EXAMPLES |
| 172 |
.SH EXAMPLES |
| 173 |
.PP |
| 174 |
The following examples use the |
| 175 |
.UR park_row.ppm |
| 176 |
park_row.ppm |
| 177 |
.UE |
| 178 |
\& test image, which is a |
| 179 |
.UR https://commons.wikimedia.org/wiki/File:15_Park_Row_3.JPG |
| 180 |
photograph of New York City's Park Row Building |
| 181 |
.UE |
| 182 |
\&, scaled to |
| 183 |
441×640, converted to a PPM file, and redistributed under the terms of |
| 184 |
the |
| 185 |
.UR https://en.wikipedia.org/wiki/GNU_Free_Documentation_License |
| 186 |
GFDL |
| 187 |
.UE |
| 188 |
\&. |
| 189 |
.PP |
| 190 |
The first example showcases the real power of bilinear transformations. |
| 191 |
Assuming \fIpark_row_rect.map\fP has the following contents: |
| 192 |
|
| 193 |
.nf\f(CW (147, 51) (316, 105) (402, 595) (92, 560) |
| 194 |
(0, 0) (440, 0) (440, 639) (0, 639)\fP |
| 195 |
.fi |
| 196 |
.PP |
| 197 |
then |
| 198 |
|
| 199 |
.nf\f(CW pamhomography -mapfile park_row_rect.map park_row.ppm > park_row_rect.ppm\fP |
| 200 |
.fi |
| 201 |
.PP |
| 202 |
projects the building's facade from a perspective view to a rectilinear |
| 203 |
front-on view. Remember that \fBpamhomography\fP ignores the parentheses and |
| 204 |
commas used in \fIpark_row_rect.map\fP; they merely make the file more |
| 205 |
human-readable. We equivalently could have written |
| 206 |
|
| 207 |
.nf\f(CW 147 51 316 105 402 595 92 560 0 0 440 0 440 639 0 639\fP |
| 208 |
.fi |
| 209 |
.PP |
| 210 |
or any of myriad other variations. |
| 211 |
.PP |
| 212 |
\fBpamhomography\fP can warp the image to a trapezoid to make it look like |
| 213 |
it's leaning backwards in 3-D: |
| 214 |
|
| 215 |
.nf\f(CW pamhomography -to '50,0 390,0 440,200 0,200' park_row.ppm > park_row_trap.ppm\fP |
| 216 |
.fi |
| 217 |
.PP |
| 218 |
As a very simple example, |
| 219 |
|
| 220 |
.nf\f(CW pamhomography -to '440,0 0,0 0,639 440,639' park_row.ppm > park_row_flip.ppm\fP |
| 221 |
.fi |
| 222 |
.PP |
| 223 |
flips the image left-to-right. Note that in this case the target |
| 224 |
quadrilateral's coordinates are listed in counterclockwise order because |
| 225 |
that represents the correspondence between points (0, 0) ↔ (440, 0) and |
| 226 |
(0, 639) ↔ (639, 0). |
| 227 |
.PP |
| 228 |
Scaling is also straightforward. The following command scales down the |
| 229 |
image from 441×640 to 341×540: |
| 230 |
|
| 231 |
.nf\f(CW pamhomography -to '0,0 340,0 340,539 0,539' park_row.ppm > park_row_small.ppm\fP |
| 232 |
.fi |
| 233 |
.PP |
| 234 |
Let's add 100 pixels of tan border to the above. We use \fB-view\fP and |
| 235 |
\fB-fill\fP to accomplish that task: |
| 236 |
|
| 237 |
.nf\f(CW pamhomography -to '0,0 340,0 340,539 0,539' -view '-100,-100 440,639' -fill tan park_row.ppm > park_row_small_border.ppm\fP |
| 238 |
.fi |
| 239 |
.PP |
| 240 |
We can add a border without having to scale the image: |
| 241 |
|
| 242 |
.nf\f(CW pamhomography -view '-100,-100 540,739' -fill tan park_row.ppm > park_row_border.ppm\fP |
| 243 |
.fi |
| 244 |
.PP |
| 245 |
The \fB-view\fP option can also be used to extract a rectangle out of an |
| 246 |
image, discarding the rest of the image: |
| 247 |
|
| 248 |
.nf\f(CW pamhomography -view '130,10 205,80' park_row.ppm > park_row_cut.ppm\fP |
| 249 |
.fi |
| 250 |
.PP |
| 251 |
Specifying the same set of coordinates to \fB-from\fP and \fB-to\fP has |
| 252 |
the same effect but also allows you to extract non-rectangular quadrilaterals |
| 253 |
from an image: |
| 254 |
|
| 255 |
.nf\f(CW pamhomography -from '185,300 310,325 320,425 180,405' -to '185,300 310,325 320,425 180,405' park_row.ppm > park_row_cut_2.ppm\fP |
| 256 |
.fi |
| 257 |
.PP |
| 258 |
Rotation is doable but takes some effort. The challenge is that you need to |
| 259 |
compute the rotated coordinates yourself. The matrix expression to rotate |
| 260 |
points \e((x_1, y_1)\e) \e((x_2, y_2)\e), \e((x_3, y_3)\e), and \e((x_4, y_4)\e) |
| 261 |
clockwise by \e(\etheta\e) degrees around point \e((c_x, c_y)\e) is |
| 262 |
.PP |
| 263 |
\e[ \ebegin{bmatrix} 1 & 0 & c_x \e\e 0 & 1 & c_y \e\e 0 & 0 |
| 264 |
& 1 \eend{bmatrix} \ebegin{bmatrix} \ecos \etheta & -\esin \etheta & 0 |
| 265 |
\e\e \esin \etheta & \ecos \etheta & 0 \e\e 0 & 0 & 1 \eend{bmatrix} |
| 266 |
\ebegin{bmatrix} 1 & 0 & -c_x \e\e 0 & 1 & -c_y \e\e 0 & 0 |
| 267 |
& 1 \eend{bmatrix} \ebegin{bmatrix} x_1 & x_2 & x_3 & x_4 \e\e y_1 |
| 268 |
& y_2 & y_3 & y_4 \e\e 1 & 1 & 1 & 1 \eend{bmatrix} |
| 269 |
\equad. \e] |
| 270 |
.PP |
| 271 |
For example, to rotate \fIpark_row.ppm\fP 30° clockwise around (220, |
| 272 |
320) you would compute |
| 273 |
.PP |
| 274 |
\e[ \ebegin{bmatrix} 1 & 0 & 220 \e\e 0 & 1 & 320 \e\e 0 & 0 |
| 275 |
& 1 \eend{bmatrix} \ebegin{bmatrix} \ecos 30^{\ecirc} & -\esin 30^{\ecirc} |
| 276 |
& 0 \e\e \esin 30^{\ecirc} & \ecos 30^{\ecirc} & 0 \e\e 0 & 0 & 1 |
| 277 |
\eend{bmatrix} \ebegin{bmatrix} 1 & 0 & -220 \e\e 0 & 1 & -320 \e\e |
| 278 |
0 & 0 & 1 \eend{bmatrix} \ebegin{bmatrix} 0 & 440 & 440 & 0 |
| 279 |
\e\e 0 & 0 & 639 & 639 \e\e 1 & 1 & 1 & 1 \eend{bmatrix} = |
| 280 |
\ebegin{bmatrix} 189.4744 & 570.5256 & 251.0256 & -130.0256 \e\e |
| 281 |
-67.1281 & 152.8719 & 706.2621 & 486.2621 \e\e 1.0000 & 1.0000 |
| 282 |
& 1.0000 & 1.0000 \eend{bmatrix} \equad, \e] |
| 283 |
.PP |
| 284 |
round these coordinates to integers, transpose the matrix, and produce the |
| 285 |
following map file, \fIpark_row_rot30.map\fP: |
| 286 |
|
| 287 |
.nf\f(CW 189 -67 |
| 288 |
571 153 |
| 289 |
251 706 |
| 290 |
-130 486\fP |
| 291 |
.fi |
| 292 |
.PP |
| 293 |
(These are the 'to' coordinates; we use the default, full-image |
| 294 |
\&'from' coordinates.) The mapping then works as in all of the |
| 295 |
preceding examples: |
| 296 |
|
| 297 |
.nf\f(CW pamhomography -mapfile park_row_rot30.map park_row.ppm > park_row_rot30.ppm\fP |
| 298 |
.fi |
| 299 |
|
| 300 |
|
| 301 |
.UN SEE-ALSO |
| 302 |
.SH SEE ALSO |
| 303 |
|
| 304 |
|
| 305 |
.IP \(bu |
| 306 |
.BR "pamcut" (1)\c |
| 307 |
\& |
| 308 |
.IP \(bu |
| 309 |
.BR "pamenlarge" (1)\c |
| 310 |
\& |
| 311 |
.IP \(bu |
| 312 |
.BR "pamflip" (1)\c |
| 313 |
\& |
| 314 |
.IP \(bu |
| 315 |
.BR "pamperspective" (1)\c |
| 316 |
\& |
| 317 |
.IP \(bu |
| 318 |
.BR "pamscale" (1)\c |
| 319 |
\& |
| 320 |
.IP \(bu |
| 321 |
.BR "pamstretch" (1)\c |
| 322 |
\& |
| 323 |
.IP \(bu |
| 324 |
.BR "pam" (1)\c |
| 325 |
\& |
| 326 |
.IP \(bu |
| 327 |
.BR "pnmmargin" (1)\c |
| 328 |
\& |
| 329 |
.IP \(bu |
| 330 |
.BR "pnmpad" (1)\c |
| 331 |
\& |
| 332 |
.IP \(bu |
| 333 |
.BR "pnmrotate" (1)\c |
| 334 |
\& |
| 335 |
.IP \(bu |
| 336 |
.BR "pnmshear" (1)\c |
| 337 |
\& |
| 338 |
|
| 339 |
|
| 340 |
|
| 341 |
.UN HISTORY |
| 342 |
.SH HISTORY |
| 343 |
.PP |
| 344 |
\fBpamhomography\fP was new in Netpbm 10.94 (March 2021). |
| 345 |
|
| 346 |
|
| 347 |
.UN AUTHOR |
| 348 |
.SH AUTHOR |
| 349 |
.PP |
| 350 |
Copyright \(co 2020 Scott |
| 351 |
Pakin, \[email protected]\fP |
| 352 |
|
| 353 |
|
| 354 |
.UN index |
| 355 |
.SH Table of Contents |
| 356 |
|
| 357 |
|
| 358 |
.IP \(bu |
| 359 |
.UR #SYNOPSIS |
| 360 |
SYNOPSIS |
| 361 |
.UE |
| 362 |
\& |
| 363 |
.IP \(bu |
| 364 |
.UR #DESCRIPTION |
| 365 |
DESCRIPTION |
| 366 |
.UE |
| 367 |
\& |
| 368 |
.IP \(bu |
| 369 |
.UR #OPTIONS |
| 370 |
OPTIONS |
| 371 |
.UE |
| 372 |
\& |
| 373 |
.IP \(bu |
| 374 |
.UR #PARAMETERS |
| 375 |
PARAMETERS |
| 376 |
.UE |
| 377 |
\& |
| 378 |
.IP \(bu |
| 379 |
.UR #NOTES |
| 380 |
NOTES |
| 381 |
.UE |
| 382 |
\& |
| 383 |
.IP \(bu |
| 384 |
.UR #EXAMPLES |
| 385 |
EXAMPLES |
| 386 |
.UE |
| 387 |
\& |
| 388 |
.IP \(bu |
| 389 |
.UR #SEE-ALSO |
| 390 |
SEE ALSO |
| 391 |
.UE |
| 392 |
\& |
| 393 |
.IP \(bu |
| 394 |
.UR #HISTORY |
| 395 |
HISTORY |
| 396 |
.UE |
| 397 |
\& |
| 398 |
.IP \(bu |
| 399 |
.UR #AUTHOR |
| 400 |
AUTHOR |
| 401 |
.UE |
| 402 |
\& |
| 403 |
.SH DOCUMENT SOURCE |
| 404 |
This manual page was generated by the Netpbm tool 'makeman' from HTML |
| 405 |
source. The master documentation is at |
| 406 |
.IP |
| 407 |
.B http://netpbm.sourceforge.net/doc/pamhomography.html |
| 408 |
.PP |