1mlib_ImageFilteredSubsamplem(e3dMiLaILBi)b Library Fumnlcitbi_oInmsageFilteredSubsample(3MLIB)
2
6 mlib_ImageFilteredSubsample, mlib_ImageFilteredSubsample_Fp - antialias
7 filters and subsamples an image
8
10 cc [ flag... ] file... -lmlib [ library... ]
11 #include <mlib.h>
12 mlib_status mlib_ImageFilteredSubsample(mlib_image *dst,
14 const mlib_image *src, mlib_s32 scaleX, mlib_s32 scaleY,
15 mlib_s32 transX, mlib_s32 transY,
16 const mlib_d64 *hKernel, const mlib_d64 *vKernel, mlib_s32 hSize,
17 mlib_s32 vSize, mlib_s32 hParity, mlib_s32 vParity, mlib_edge edge);
18 mlib_status mlib_ImageFilteredSubsample_Fp(mlib_image *dst,
21 const mlib_image *src, mlib_s32 scaleX, mlib_s32 scaleY,
22 mlib_s32 transX, mlib_s32 transY,
23 const mlib_d64 *hKernel, const mlib_d64 *vKernel, mlib_s32 hSize,
24 mlib_s32 vSize, mlib_s32 hParity, mlib_s32 vParity, mlib_edge edge);
25
28 Each of the functions antialias filters and subsamples an image.
29 The effect of one of the functions on an image is equivalent to per‐
32 forming convolution (filter) followed by subsampling (zoom out).
33 The functions are similar to the mlib_ImageZoomTranslate() and
36 mlib_ImageZoomTranslate_Fp() functions. But they have different defini‐
37 tions on scale factors and translations, hence use different coordinate
38 mapping equations. The scaleX and scaleY used by mlib_ImageFilteredSub‐
39 sample() and mlib_ImageFilteredSubsample_Fp() are the reciprocals of
40 the zoomx and zoomy, respectively, used by mlib_ImageZoomTranslate()
41 and mlib_ImageZoomTranslate_Fp().
42 The functions use the following equations for coordinate mapping:
45 xS = xD*scaleX + transX
47 yS = yD*scaleY + transY
48 where, a point (xD, yD) in the destination image is backward mapped to
52 a point (xS, yS) in the source image. The arguments transX and transY
53 are provided to support tiling.
54 The subsample terms, i.e., the scale factors scaleX and scaleY, are
57 restricted to positive integral values. Geometrically, one destination
58 pixel maps to scaleX by scaleY source pixels. With odd scale factors,
59 destination pixel centers map directly onto source pixel centers. With
60 even scale factors, destination pixel centers map squarely between
61 source pixel centers. Below are examples of even, odd, and combination
62 cases.
63 s s s s s s s s s s s s
65 d d d
66 s s s s s s s d s s d s
67 s s s s s s s s s s s s
69 d d d
70 s s s s s s s s s s s s
71 s s s s s s s d s s d s
73 d d d
74 s s s s s s s s s s s s
75 Even scaleX/Y factors Odd scaleX/Y factors
77 s s s s s s s s s s s s
79 d d
80 s s s s s s s d s s d s s d s
81 s s s s s s s s s s s s
83 d d
84 s s s s s s s s s s s s
85 s s s s s s s d s s d s s d s
87 d d
88 s s s s s s s s s s s s
89 Odd/even scaleX/Y factors Even/odd scaleX/Y factors
91 where
95 s = source pixel centers
97 d = destination pixel centers mapped to source
98 The applied filter is quadrant symmetric (typically antialias + resam‐
102 ple). The filter is product-separable, quadrant symmetric, and is
103 defined by half of its span. Parity is used to signify whether the sym‐
104 metric kernel has a double center (even parity) or a single center
105 value (odd parity). For example, if hParity == 0 (even), the horizontal
106 kernel is defined as:
107 hKernel[hSize-1], ..., hKernel[0], hKernel[0], ...,
109 hKernel[hSize-1]
110 Otherwise, if hParity == 1 (odd), the horizontal kernel is defined as:
114 hKernel[hSize-1], ..., hKernel[0], ...,
116 hKernel[hSize-1]
117 Horizontal and vertical kernels representing convolved resample (i.e.,
121 the combined separable kernels) can be computed from a convolution fil‐
122 ter (with odd parity), a resample filter, and because the subsample
123 factors affect resample weights, the subsample scale factors. It is the
124 user's responsibility to provide meaningful combined kernels.
125 To compute the value of a pixel centered at point (xD, yD) in the des‐
128 tination image, apply the combined kernel to the source image by align‐
129 ing the kernel's geometric center to the backward mapped point (xS, yS)
130 in the source image. In the cases that it can not be exactly on top of
131 point (xS, yS), the kernel's center should be half-pixel right and/or
132 below that point. When this is done in a separable manner, the centers
133 of horizontal and vertical kernels should align with xS and yS, respec‐
134 tively.
135 The combination of subsampling and filtering has performance benefits
138 over sequential fucntion usage in part due to the symmetry constraints
139 imposed by only allowing integer parameters for scaling and only allow‐
140 ing separable symmetric filters.
141
143 The function takes the following arguments:
144 dst Pointer to destination image.
146 src Pointer to source image.
149 scaleX The x scale factor of subsampling.
152 scaleY The y scale factor of subsampling.
155 transX The x translation.
158 transY The y translation.
161 hKernel Pointer to the compact form of horizontal kernel.
164 vKernel Pointer to the compact form of vertical kernel.
167 hSize Size of array hKernel.
170 vSize Size of array vKernel.
173 hParity Parity of horizontal kernel (0: even, 1: odd).
176 vParity Parity of vertical kernel (0: even, 1: odd).
179 edge Type of edge condition. It can be one of the following:
182 MLIB_EDGE_DST_NO_WRITE
184
188 The function returns MLIB_SUCCESS if successful. Otherwise it returns
189 MLIB_FAILURE.
190
192 See attributes(5) for descriptions of the following attributes:
193 ┌─────────────────────────────┬─────────────────────────────┐
198 │ ATTRIBUTE TYPE │ ATTRIBUTE VALUE │
199 ├─────────────────────────────┼─────────────────────────────┤
200 │Interface Stability │Committed │
201 ├─────────────────────────────┼─────────────────────────────┤
202 │MT-Level │MT-Safe │
203 └─────────────────────────────┴─────────────────────────────┘
204
206 mlib_ImageSubsampleAverage(3MLIB), mlib_ImageZoomTranslate(3MLIB),
207 mlib_ImageZoomTranslate_Fp(3MLIB), attributes(5)
208SunOS 5.11 2 Mar 2007mlib_ImageFilteredSubsample(3MLIB)