1
\documentclass{article}
5
\title{Some image transform math}
7
\date{18 February 2003}
12
The transform process is composed of three steps;
13
first we reconstruct a continuous image from the
14
source data \(A_{i,j}\):
15
\[a(u,v) = \sum_{i = -\infty}^{\infty} \sum_{j = -\infty}^{\infty} A_{i,j}F\left( {u - i \atop v - j} \right) \]
16
Then we transform from destination coordinates to source coordinates:
17
\[b(x,y) = a\left(u(x,y) \atop v(x,y)\right)
18
= a\left(t_{00}x + t_{01}y + t_{02} \atop t_{10}x + t_{11}y + t_{12} \right)\]
19
Finally, we resample using a sampling function \(G\):
20
\[B_{x_0,y_0} = \int_{-\infty}^{\infty}\int_{-\infty}^{\infty} b(x,y)G\left( {x - x_0 \atop y - y_0} \right) dxdy\]
21
Putting all of these together:
23
\int_{-\infty}^{\infty}\int_{-\infty}^{\infty}
24
\sum_{i = -\infty}^{\infty} \sum_{j = -\infty}^{\infty} A_{i,j}
25
F\left( {u(x,y) - i \atop v(x,y) - j} \right)
26
G\left( {x - x_0 \atop y - y_0} \right) dxdy\]
27
We can reverse the order of the integrals and the sums:
29
\sum_{i = -\infty}^{\infty} \sum_{j = -\infty}^{\infty} A_{i,j}
30
\int_{-\infty}^{\infty}\int_{-\infty}^{\infty}
31
F\left( {u(x,y) - i \atop v(x,y) - j} \right)
32
G\left( {x - x_0 \atop y - y_0} \right) dxdy\]
33
Which shows that the destination pixel values are a linear combination of the
34
source pixel values. But the coefficents depend on \(x_0\) and \(y_0\).
35
To simplify this a bit, define:
36
\[i_0 = \lfloor u(x_0,y_0) \rfloor = \lfloor {t_{00}x_0 + t_{01}y_0 + t_{02}} \rfloor \]
37
\[j_0 = \lfloor v(x_0,y_0) \rfloor = \lfloor {t_{10}x_0 + t_{11}y_0 + t_{12}} \rfloor \]
38
\[\Delta_u = u(x_0,y_0) - i_0 = t_{00}x_0 + t_{01}y_0 + t_{02} - \lfloor {t_{00}x_0 + t_{01}y_0 + t_{02}} \rfloor \]
39
\[\Delta_v = v(x_0,y_0) - j_0 = t_{10}x_0 + t_{11}y_0 + t_{12} - \lfloor {t_{10}x_0 + t_{11}y_0 + t_{12}} \rfloor \]
40
Then making the transforms \(x' = x - x_0\), \(y' = y - x_0\), \(i' = i - i_0\), \(j' = j - x_0\)
42
F(u,v) & = & F\left( {t_{00}x + t_{01}y + t_{02} - i \atop t_{10}x + t_{11}y + t_{12} - j} \right)\\
43
& = & F\left( {t_{00}(x'+x_0) + t_{01}(y'+y_0) + t_{02} - (i'+i_0) \atop
44
t_{10}(x'+x_0) + t_{11}(y'+y_0) + t_{12} - (j'+j_0)} \right) \\
45
& = & F\left( {\Delta_u + t_{00}x' + t_{01}y' - i' \atop
46
\Delta_v + t_{10}x' + t_{11}y' - j'} \right)
48
Using that, we can then reparameterize the sums and integrals and
49
define coefficients that depend only on \((\Delta_u,\Delta_v)\),
50
which we'll call the \emph{phase} at the point \((x_0,y_0)\):
53
\sum_{i = -\infty}^{\infty} \sum_{j = -\infty}^{\infty} A_{i_0+i,j_0+j} C_{i,j}(\Delta_u,\Delta_v)
56
C_{i,j}(\Delta_u,\Delta_v) =
57
\int_{-\infty}^{\infty}\int_{-\infty}^{\infty}
58
F\left( {\Delta_u + t_{00}x + t_{01}y - i \atop
59
\Delta_v + t_{10}x + t_{11}y - j} \right)
60
G\left( {x \atop y} \right) dxdy
62
\section{Separability}
63
A frequent special case is when the reconstruction and sampling functions
67
If we also have a transform that is purely a scale and translation;
68
(\(t_{10} = 0\), \(t_{01} = 0\)), then we can separate
69
\(C_{i,j}(\Delta_u,\Delta_v)\) into the product of a \(x\) portion
71
\[C_{i,j}(\Delta_u,\Delta_v) = c_{i}(\Delta_u) c_{j}(\Delta_v)\]
72
\[c_{i}(\Delta_u) = \int_{-\infty}^{\infty} f(\Delta_u + t_{00}x - i)g(x)dx\]
73
\[c_{j}(\Delta_v) = \int_{-\infty}^{\infty} f(\Delta_v + t_{11}y - j)g(y)dy\]
75
\section{Some filters}
76
gdk-pixbuf provides 4 standard filters for scaling, under the names ``NEAREST'',
77
``TILES'', ``BILINEAR'', and ``HYPER''. All of turn out to be separable
78
as discussed in the previous section.
79
For ``NEAREST'' filter, the reconstruction function is simple replication
80
and the sampling function is a delta function\footnote{A delta function is an infinitely narrow spike, such that:
81
\[\int_{-\infty}^{\infty}\delta(x)f(x) = f(0)\]}:
82
\[f(t) = \cases{1, & if \(0 \le t \le 1\); \cr
84
\[g(t) = \delta(t - 0.5)\]
85
For ``TILES'', the reconstruction function is again replication, but we
86
replace the delta-function for sampling with a box filter:
87
\[f(t) = \cases{1, & if \(0 \le t \le 1\); \cr
89
\[g(t) = \cases{1, & if \(0 \le t \le 1\); \cr
91
The ``HYPER'' filter (in practice, it was originally intended to be
92
something else) uses bilinear interpolation for reconstruction and
93
a box filter for sampling:
94
\[f(t) = \cases{1 - |t - 0.5|, & if \(-0.5 \le t \le 1.5\); \cr
96
\[g(t) = \cases{1, & if \(0 \le t \le 1\); \cr
98
The ``BILINEAR'' filter is defined in a somewhat more complicated way;
99
the definition depends on the scale factor in the transform (\(t_{00}\)
100
or \(t_{01}]\). In the \(x\) direction, for \(t_{00} < 1\), it is
101
the same as for ``TILES'':
102
\[f_u(t) = \cases{1, & if \(0 \le t \le 1\); \cr
104
\[g_u(t) = \cases{1, & if \(0 \le t \le 1\); \cr
106
but for \(t_{10} > 1\), we use bilinear reconstruction and delta-function
108
\[f_u(t) = \cases{1 - |t - 0.5|, & if \(-0.5 \le t \le 1.5\); \cr
110
\[g_u(t) = \delta(t - 0.5)\]
111
The behavior in the \(y\) direction depends in the same way on \(t_{11}\).
b'\\ No newline at end of file'