Penguin
Recent Changes
Blame: pnmnlfilt(1)
EditPageHistoryDiffInfoLikePages
Annotated edit history of pnmnlfilt(1) version 1, including all changes. View license author blame.
Rev Author # Line
1 perry 1 pnmnlfilt
2 !!!pnmnlfilt
3 NAME
4 SYNOPSIS
5 DESCRIPTION
6 Alpha trimmed mean filter. (0.0
7 Optimal estimation smoothing. (1.0
8 Edge enhancement. (-0.1
9 Combination use.
10 References:
11 SEE ALSO
12 BUGS
13 AUTHOR
14 ----
15 !!NAME
16
17
18 pnmnlfilt - non-linear filters: smooth, alpha trim mean, optimal estimation smoothing, edge enhancement.
19 !!SYNOPSIS
20
21
22 __pnmnlfilt__ alpha radius [[''pnmfile'']
23 !!DESCRIPTION
24
25
26 This is something of a swiss army knife filter. It has 3
27 distinct operating modes. In all of the modes each pixel in
28 the image is examined and processed according to it and its
29 surrounding pixels values. Rather than using the 9 pixels in
30 a 3x3 block, 7 hexagonal area samples are taken, the size of
31 the hexagons being controlled by the radius parameter. A
32 radius value of 0.3333 means that the 7 hexagons exactly fit
33 into the center pixel (ie. there will be no filtering
34 effect). A radius value of 1.0 means that the 7 hexagons
35 exactly fit a 3x3 pixel array.
36 !!Alpha trimmed mean filter. (0.0 !!
37
38
39 The value of the center pixel will be replaced by the mean
40 of the 7 hexagon values, but the 7 values are sorted by size
41 and the top and bottom alpha portion of the 7 are excluded
42 from the mean. This implies that an alpha value of 0.0 gives
43 the same sort of output as a normal convolution (ie.
44 averaging or smoothing filter), where radius will determine
45 the
46
47
48 An alpha value of 0.5 will cause the median value of the 7
49 hexagons to be used to replace the center pixel value. This
50 sort of filter is good for eliminating
51 !!Optimal estimation smoothing. (1.0 !!
52
53
54 This type of filter applies a smoothing filter adaptively
55 over the image. For each pixel the variance of the
56 surrounding hexagon values is calculated, and the amount of
57 smoothing is made inversely proportional to it. The idea is
58 that if the variance is small then it is due to noise in the
59 image, while if the variance is large, it is because of
60 !!Edge enhancement. (-0.1 !!
61
62
63 This is the opposite type of filter to the smoothing filter.
64 It enhances edges. The alpha parameter controls the amount
65 of edge enhancement, from subtle (-0.1) to blatant (-0.9).
66 The radius parameter controls the effective radius as usual,
67 but useful values are between 0.5 and 0.9. Try starting with
68 values of alpha = 0.3, radius = 0.8
69 !!Combination use.
70
71
72 The various modes of __pnmnlfilt__ can be used one after
73 the other to get the desired result. For instance to turn a
74 monochrome dithered image into a grayscale image you could
75 try one or two passes of the smoothing filter, followed by a
76 pass of the optimal estimation filter, then some subtle edge
77 enhancement. Note that using edge enhancement is only likely
78 to be useful after one of the non-linear filters (alpha
79 trimmed mean or optimal estimation filter), as edge
80 enhancement is the direct opposite of
81 smoothing.
82
83
84 For reducing color quantization noise in images (ie. turning
85 .gif files back into 24 bit files) you could try a pass of
86 the optimal estimation filter (alpha 1.2, radius 1.0), a
87 pass of the median filter (alpha 0.5, radius 0.55), and
88 possibly a pass of the edge enhancement filter. Several
89 passes of the optimal estimation filter with declining alpha
90 values are more effective than a single pass with a large
91 alpha value. As usual, there is a tradeoff between filtering
92 effectiveness and loosing detail. Experimentation is
93 encouraged.
94 !!References:
95
96
97 The alpha-trimmed mean filter is based on the description in
98 IEEE CG
99
100
101 The optimal estimation filter is taken from an article
102
103
104 The edge enhancement details are from pgmenhance(1), which
105 is taken from Philip R. Thompson's
106 !!SEE ALSO
107
108
109 pgmenhance(1), pnmconvol(1), pnm(5)
110 !!BUGS
111
112
113 Integers and tables may overflow if PPM_MAXMAXVAL is greater
114 than 255.
115 !!AUTHOR
116
117
118 Graeme W. Gill graeme@labtam.oz.au
119 ----
This page is a man page (or other imported legacy content). We are unable to automatically determine the license status of this page.