1ImageND(3) User Contributed Perl Documentation ImageND(3)
2
6 PDL::ImageND - useful image processing in N dimensions
7
9 These routines act on PDLs as N-dimensional objects, not as threaded
10 sets of 0-D or 1-D objects. The file is sort of a catch-all for
11 broadly functional routines, most of which could legitimately be filed
12 elsewhere (and probably will, one day).
13 ImageND is not a part of the PDL core (v2.4) and hence must be explic‐
15 itly loaded.
16
18 use PDL::ImageND;
19 $b = $a->convolveND($kernel,{bound=>'periodic'});
21 $b = $a->rebin(50,30,10);
22
24 convolve
25 Signature: (a(m); b(n); int adims(p); int bdims(q); [o]c(m))
27 N-dimensional convolution (Deprecated; use convolveND)
29 $new = convolve $a, $kernel
31 Convolve an array with a kernel, both of which are N-dimensional. This
33 routine does direct convolution (by copying) but uses quasi-periodic
34 boundary conditions: each dim "wraps around" to the next higher row in
35 the next dim.
36 This routine is kept for backwards compatibility with earlier scripts;
38 for most purposes you want convolveND instead: it runs faster and han‐
39 dles a variety of boundary conditions.
40 ninterpol()
42 N-dimensional interpolation routine
44 Signature: ninterpol(point(),data(n),[o]value())
46 $value = ninterpol($point, $data);
48 "ninterpol" uses "interpol" to find a linearly interpolated value in N
50 dimensions, assuming the data is spread on a uniform grid. To use an
51 arbitrary grid distribution, need to find the grid-space point from the
52 indexing scheme, then call "ninterpol" -- this is far from trivial (and
53 ill-defined in general).
54 See also interpND, which includes boundary conditions and allows you to
56 switch the method of interpolation, but which runs somewhat slower.
57 rebin
59 Signature: (a(m); [o]b(n); int ns => n)
61 N-dimensional rebinning algorithm
63 $new = rebin $a, $dim1, $dim2,..;. $new = rebin $a, $template; $new =
65 rebin $a, $template, {Norm => 1};
66 Rebin an N-dimensional array to newly specified dimensions. Specifying
68 `Norm' keeps the sum constant, otherwise the intensities are kept con‐
69 stant. If more template dimensions are given than for the input pdl,
70 these dimensions are created; if less, the final dimensions are main‐
71 tained as they were.
72 So if $a is a 10 x 10 pdl, then "rebin($a,15)" is a 15 x 10 pdl, while
74 "rebin($a,15,16,17)" is a 15 x 16 x 17 pdl (where the values along the
75 final dimension are all identical).
76 Expansion is performed by sampling; reduction is performed by averag‐
78 ing. If you want different behavior, use PDL::Transform::map instead.
79 PDL::Transform::map runs slower but is more flexible.
80 circ_mean_p
82 Calculates the circular mean of an n-dim image and returns the projec‐
84 tion. Optionally takes the center to be used.
85 $cmean=circ_mean_p($im);
87 $cmean=circ_mean_p($im,{Center => [10,10]});
88 circ_mean
90 Smooths an image by applying circular mean. Optionally takes the cen‐
92 ter to be used.
93 circ_mean($im);
95 circ_mean($im,{Center => [10,10]});
96 kernctr()
98 `centre' a kernel (auxiliary routine to fftconvolve)
100 $kernel = kernctr($image,$smallk);
102 fftconvolve($image,$kernel);
103 kernctr centres a small kernel to emulate the behaviour of the direct
105 convolution routines.
106 convolveND
108 Signature: (k0(); SV *k; SV *aa; SV *a)
110 Speed-optimized convolution with selectable boundary conditions
112 $new = convolveND($a, $kernel, [ {options} ]);
114 Conolve an array with a kernel, both of which are N-dimensional.
116 If the kernel has fewer dimensions than the array, then the extra array
118 dimensions are threaded over. There are options that control the
119 boundary conditions and method used.
120 The kernel's origin is taken to be at the kernel's center. If your
122 kernel has a dimension of even order then the origin's coordinates get
123 rounded up to the next higher pixel (e.g. (1,2) for a 3x4 kernel).
124 This mimics the behavior of the earlier convolve and fftconvolve rou‐
125 tines, so convolveND is a drop-in replacement for them.
126 The kernel may be any size compared to the image, in any dimension.
128 The kernel and the array are not quite interchangeable (as in mathemat‐
130 ical convolution): the code is inplace-aware only for the array itself,
131 and the only allowed boundary condition on the kernel is truncation.
132 convolveND is inplace-aware: say "convolveND(inplace $a ,$k)" to modify
134 a variable in-place. You don't reduce the working memory that way --
135 only the final memory.
136 OPTIONS
138 Options are parsed by PDL::Options, so unique abbreviations are
140 accepted.
141 boundary (default: 'truncate')
143 The boundary condition on the array, which affects any pixel closer
144 to the edge than the half-width of the kernel.
145 The boundary conditions are the same as those accepted by range,
147 because this option is passed directly into range. Useful options
148 are 'truncate' (the default), 'extend', and 'periodic'. You can
149 select different boundary conditions for different axes -- see range
150 for more detail.
151 The (default) truncate option marks all the near-boundary pixels as
153 BAD if you have bad values compiled into your PDL.
154 method (default: 'auto')
156 The method to use for the convolution. Acceptable alternatives are
157 'direct', 'fft', or 'auto'. The direct method is an explicit copy-
158 and-multiply operation; the fft method takes the Fourier transform
159 of the input and output kernels. The two methods give the same
160 answer to within double-precision numerical roundoff. The fft
161 method is much faster for large kernels; the direct method is faster
162 for tiny kernels. The tradeoff occurs when the array has about 400x
163 more pixels than the kernel.
164 The default method is 'auto', which chooses direct or fft convolu‐
166 tion based on the size of the input arrays.
167 NOTES
169 At the moment there's no way to thread over kernels. That could/should
171 be fixed.
172 The threading over input is cheesy and should probably be fixed: cur‐
174 rently the kernel just gets dummy dimensions added to it to match the
175 input dims. That does the right thing tersely but probably runs slower
176 than a dedicated threadloop.
177 The direct copying code uses PP primarily for the generic typing: it
179 includes its own threadloops.
180
182 Copyright (C) Karl Glazebrook and Craig DeForest, 1997, 2003 All rights
183 reserved. There is no warranty. You are allowed to redistribute this
184 software / documentation under certain conditions. For details, see the
185 file COPYING in the PDL distribution. If this file is separated from
186 the PDL distribution, the copyright notice should be included in the
187 file.
188perl v5.8.8 2006-12-02 ImageND(3)