kopia lustrzana https://gitlab.com/sane-project/website
119 wiersze
5.9 KiB
HTML
119 wiersze
5.9 KiB
HTML
<!-- received="Mon Sep 1 14:28:23 1997 MST" -->
|
|
<!-- sent="Mon, 1 Sep 1997 22:37:36 +0200 (MET DST)" -->
|
|
<!-- name="Rogier Wolff" -->
|
|
<!-- email="R.E.Wolff@BitWizard.nl" -->
|
|
<!-- subject="Re: umax-backend" -->
|
|
<!-- id="199709012037.WAA00913@cave.BitWizard.nl" -->
|
|
<!-- inreplyto="m0x5VrW-000J8HC@hex.prestel.co.uk" -->
|
|
<title>sane-devel: Re: umax-backend</title>
|
|
<h1>Re: umax-backend</h1>
|
|
<b>Rogier Wolff</b> (<a href="mailto:R.E.Wolff@BitWizard.nl"><i>R.E.Wolff@BitWizard.nl</i></a>)<br>
|
|
<i>Mon, 1 Sep 1997 22:37:36 +0200 (MET DST)</i>
|
|
<p>
|
|
<ul>
|
|
<li> <b>Messages sorted by:</b> <a href="date.html#121">[ date ]</a><a href="index.html#121">[ thread ]</a><a href="subject.html#121">[ subject ]</a><a href="author.html#121">[ author ]</a>
|
|
<!-- next="start" -->
|
|
<li> <b>Next message:</b> <a href="0122.html">Jonathan A. Buzzard: "Re: umax-backend"</a>
|
|
<li> <b>Previous message:</b> <a href="0120.html">Oliver.Rauch@Wolfsburg.DE: "umax-backend"</a>
|
|
<li> <b>In reply to:</b> <a href="0119.html">Jonathan A. Buzzard: "Re: umax-backend"</a>
|
|
<!-- nextthread="start" -->
|
|
<li> <b>Next in thread:</b> <a href="0122.html">Jonathan A. Buzzard: "Re: umax-backend"</a>
|
|
<!-- reply="end" -->
|
|
</ul>
|
|
<!-- body="start" -->
|
|
Jonathan A. Buzzard wrote:<br>
|
|
<i>> </i><br>
|
|
<i>> </i><br>
|
|
<i>> <a href="mailto:R.E.Wolff@BitWizard.nl">R.E.Wolff@BitWizard.nl</a> said:</i><br>
|
|
<i>> > Yes, the q/a program will show "only noise" still, the doctors won't </i><br>
|
|
<i>> > allow you to trhow away that information. Maybe if you average 4 </i><br>
|
|
<i>> > pixels together that you get a little information. This IS visible to </i><br>
|
|
<i>> > the naked eye..... </i><br>
|
|
<i>> </i><br>
|
|
<i>> Never said the radiologist would let you throw the info away, but noise</i><br>
|
|
<i>> is noise and averaging bits together will (should)</i><br>
|
|
<p>
|
|
should not make a difference. Right. <br>
|
|
<p>
|
|
<i>> > How do you measure the "just noise" in a 12bit scanner? I'd try to </i><br>
|
|
<i>> > generate a perfect analog greyscale and scan that.</i><br>
|
|
<i>> </i><br>
|
|
<i>> Not that I was involved in the q/a program, (it was an M.Sc. project done</i><br>
|
|
<i>> at the same time as mime, and I had to use the same scanner for some</i><br>
|
|
<i>> radiotherapy films). Roughly their was a test film which had a large number</i><br>
|
|
<i>> of squares of different (but uniform) optical density. Roughly a scan was</i><br>
|
|
<i>> taken of the film, a template was placed over the image and some statistics</i><br>
|
|
<i>> where done on the pixels of the different areas. As far as I understood</i><br>
|
|
<i>> the analysis showed that on a uniform area the distribution of the values</i><br>
|
|
<i>> in the lower two bits was more or less uniform. The conclusions drawn from</i><br>
|
|
<i>> that it must be noise.</i><br>
|
|
<p>
|
|
If "that is all there is to it", then I don't think this is a valid <br>
|
|
conclusion. I think I'd have to read the whole thesis. <br>
|
|
<p>
|
|
My explanation of the "uniformity" of the values in the lower bits is<br>
|
|
"there is noise in the source image". <br>
|
|
<p>
|
|
Yeah, it may look VERY uniform, but we're trying to detect the noise<br>
|
|
in the area that we're agreeing on that humans (almost) don't see the<br>
|
|
difference any more. <br>
|
|
<p>
|
|
Try creating a greyscale image: example:<br>
|
|
position -> 1234556789<br>
|
|
/ 1234556789<br>
|
|
_/ 1234556789<br>
|
|
^ / 1234556789<br>
|
|
intensity| / 1234556789<br>
|
|
<p>
|
|
and notice the white stripe. <br>
|
|
<p>
|
|
% xv -perfect -geometry 260x240 -<br>
|
|
P2 <br>
|
|
260 1 255<br>
|
|
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26<br>
|
|
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49<br>
|
|
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72<br>
|
|
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95<br>
|
|
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113<br>
|
|
114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 128 128<br>
|
|
128 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143<br>
|
|
144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160<br>
|
|
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177<br>
|
|
178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194<br>
|
|
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211<br>
|
|
212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228<br>
|
|
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245<br>
|
|
246 247 248 249 250 251 252 253 254 255<br>
|
|
^D (EOF)<br>
|
|
<p>
|
|
The only thing I changed from the ramp is that the value 128 is <br>
|
|
mentioned 5 times. The human eye makes that into a WHITE band. <br>
|
|
<p>
|
|
(This effect requires 8 bit deep DAC's.)<br>
|
|
<p>
|
|
What might be a reasonable test is to scan the same image over and<br>
|
|
over again. For every position in the image you can then calculate<br>
|
|
the statistical parameters for that position. The standard deviation<br>
|
|
has something to do with the amount of noise. I expect the pixels<br>
|
|
on sharp edges to show a higher noise figure than other pixels<br>
|
|
due to mechanical positioning problems. <br>
|
|
<p>
|
|
Roger. <br>
|
|
<p>
|
|
<pre>
|
|
--
|
|
Source code, list archive, and docs: <a href="http://www.azstarnet.com/~axplinux/sane/">http://www.azstarnet.com/~axplinux/sane/</a>
|
|
To unsubscribe: mail -s unsubscribe <a href="mailto:sane-devel-request@listserv.azstarnet.com">sane-devel-request@listserv.azstarnet.com</a>
|
|
</pre>
|
|
<!-- body="end" -->
|
|
<p>
|
|
<ul>
|
|
<!-- next="start" -->
|
|
<li> <b>Next message:</b> <a href="0122.html">Jonathan A. Buzzard: "Re: umax-backend"</a>
|
|
<li> <b>Previous message:</b> <a href="0120.html">Oliver.Rauch@Wolfsburg.DE: "umax-backend"</a>
|
|
<li> <b>In reply to:</b> <a href="0119.html">Jonathan A. Buzzard: "Re: umax-backend"</a>
|
|
<!-- nextthread="start" -->
|
|
<li> <b>Next in thread:</b> <a href="0122.html">Jonathan A. Buzzard: "Re: umax-backend"</a>
|
|
<!-- reply="end" -->
|
|
</ul>
|