Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.


Go Back   Doom9's Forum > Capturing and Editing Video > Avisynth Usage

Thread Tools Search this Thread Display Modes
Old 30th November 2004, 21:50   #1  |  Link
Registered User
Join Date: Oct 2001
Posts: 15
waveletreducer & wavelet2DNR. Anyone understands Japanese?

Looking for wavelet denoisers, I found this one.
Too bad I don't read Japanese.
Anyone care to traduce the help file?


Last edited by oledoe; 30th November 2004 at 21:52.
oledoe is offline   Reply With Quote
Old 1st December 2004, 13:05   #2  |  Link
Registered User
Join Date: Oct 2001
Posts: 15
I translated the helpfiles using Amikai

I translated them using

The results are:



WaveletReducer 0.4.1 By Mr.Ty
WaveletReducer.DLL is the aggregate using Wavelet of NR function.
Although the author is developing / testing based on Avisynth2.0x, the thing once compiled by the header for 2.5x is also attached.
It is [ the following and ] explanation of a rough function.

waveletReducer : It is NR filter using Wavelet degeneration.
Wavelet degeneration In order to cut and reconstruct the energy considered to be a noise State which required the film with thin NR result like usual 2D NR には -- it is hard to become and becomes easy to be finished on the screen felt more refreshed, However, an outline will be ruined if side effects take a large 閾 value by slight size.
wavelet2D NR : It is 2D NR using the sub band decomposition by Wavelet conversion.
The difference from other 2D NR is the point that a 閾 value can be adjusted for every frequency.
As [ say / that color MURA of 4Dot units is taken without breaking down the pattern of 1Dot unit ]
It is helpful to a case.

Wavelet degeneration was used although there were also few side effects. In NR, it was different, the noise signal has only melted into the surroundings, and it is in a noise. Since it remains as it is, as for the increased energy, a thin film completion It becomes easy to become textures which covered the whole.

Detailed explanation of each function should look at each document (waveletReducer.txt, wavelet2DNR.txt).

Travels :

2003.03.16: Ver0.4.1 :
the place investigated in response to the report that there is no object for 2.5x in skillful _動 -- somehow -- sauce -- a part -- The alteration corrected by necessity. The thing which release without testing, which will be dispelled and to say Happening is the grade which test environment cannot prepare satisfactorily although it understands. Please understand the situation. The maximum level value of wavelet2D NR was changed into 4. と -- or it says -- the last version -- level 5 the value of something that is on some memory by the judgment of threshold6 when it carries out -- referring to -- suitable -- it is said that NR will be hung -- it was fearful This differs from waveletReducer. It is what happens in order to perform NR also in the time of a level 0 using threshold1. threshold 1-5 is a level. Since it used up by 4, the maximum level value was set to 4. Increasing one threshold, a level is 5. Although possibly there was also a method of as, since I felt that there was nothing a meaning not much.

2003. 03.15;Ver0.4.0 :
When performing Wavelet degeneration, it is the kind (whether it says, length) of Mother wavelet. It turns out that it is necessary to change restriction of how far to NR. It is an incoming signal in fact. It is necessary to protect the part from the last element to the length of the Mother wave let / 2-1. And it changes such. With the old way, they are Mother wave let other than type0.type1. It was lost although the interference wave had occurred in the screen left and the top when chosen. type5 was added to the kind of Mother wave let. It is a 閾 value, so that a type value is large. When it raises, it becomes easy to prevent an outline becoming square. Since が and calculation time increase It becomes heavy. The version was set to 0.4.0 in accordance with the wavelet2DNR function addition. The waveletReducer itself It has not escaped from the research stage yet.

2003.03.08: Ver0.3.5 :
In case the result which carried out Wavelet reverse conversion is returned as an image, 1 figure is rounded off below a decimal point. Having forgotten is revealed and it corrects confusedly. Block noise generated in this cause I think that it will be. The binary for avisynth2.5 was also appended. However, my PC for development is fundamentally. Since there is only one set and Avisynth2.07 is put in usually, move by the object for Avisynth2.5 perfectly. It cannot check in how. Those who use by the reason for saying need your help by accountability.

2003.03.06: Ver0.3.4 :
When the direction of Y has a large level number setup, the bug to which the image was confused is corrected. Work domain which must be initialized before carrying out NR of the direction of length at mode 3:00 The bug which was not initialized is corrected. The bug which was carrying out the memory leak of the 2kByte(s) at the time of class deletion is corrected after class creation. mode4 is established newly. It is whether the roughness of an outline became MASHI to some extent by this. Although considered, how is it? if the Mother wave let is set to type4 etc. Since there will be a total of four Wavelet deployment values which affect a certain block since a result becomes complicated and mysterious also as for simple operation, while expecting a result -- program carrying out -- inside ?X -- it is difficult work Although it is continuation of trial and error, it is .... somehow or other.

2003.03.03: Ver0.3.3 :
It stopped having carried out Wavelet deployment lengthwise to the horizontal Wavelet deployment value. For this, a rabbit will also be an angle, if it uses for picture compression semantically, although it becomes slanting Wavelet deployment. I felt that there was nothing in a noise filter a meaning, and a slanting Wavelet deployment value is received above all. Since I felt that NR had produced the present condition of being unusually weak, to the slanting line, it cannot take. It stopped. Since a vertical Wavelet deployment speed doubles by this, it is also a speed rise certainly. It should be connected. Generally, it does and ないっぽい and Wavelet deployment slanting also at certain Type-G is ?B.

2003.03.03: Ver0.3.2 :
The procession hung by mode3 is changed into the procession of 3*3 from the procession of 5*5. While the checks of edge are as few as the effect which was too severe and which eases it since like I think that a speed rise is expectable. It is a open start to 2ch(es). Now, thing which what denunciation and influence come ....

2003.03.02:Ver0.3.1 :
mode3 establishment. Although it is considering as middle upgrade whenever mode increases Probably, there is no it for a while from now on. While mode is for a while, it thinks that it is the close in three. A vertical Wavelet deployment value is that only root 2 are large after deployment of this horizontal level. Vertical NR is that forget and only 1/root 2 had stopped being able to be effective easily from horizontal NR until now. It corrected.

2003.02.23: Ver0.2.1 :
The state of a screen end was changed so that the data in front of a screen end might be repeated. Thereby, the outline roughness which had come out at the screen end may be improvable. mode2 was established newly. Although it is Wavelet degeneration by Soft Thresholding The NR effect is appearance [ raise ] too. However, it is perfunctorily about a parameter setup. It will carry out, or it becomes, and since it will be severe, cautions are required. usual [ BOKA / for the time being / if what relation for two points is / still / usual ] what is better than 2D NR -- かなぁ -- ? -- I consider Since threshold stopped calling NR processing to the level of 0, it is threshold1. When it was 0, possibly it has accelerated comparatively effectively. It is w which became 0 recommendation by the reason for saying unless especially threshold1 had the reason. It was made not to NR to the screen end. Even if it repeats the data in front of a screen end Although it is the thing [ like ] which had still detected edge at the screen end, mere a few is also now. If it becomes MASHI ....

2003.02.21:Ver0.1.2 :
The portion which was being calculated by double is changed into float. It accelerates merely slightly. However, the possibility ant of error accumulation. It corrects having processed by the bug to the portion which must not be processed. It corrected, although processing of a screen end must have been amusing until now. 768 was added to the breadth size after regular-izing. This Grade improvement in the speed which has received the input of 704 and the breadth size of 720 was attained. although eight or more levels are not made since it is not the multiplier of 2 -- primarily -- Since it does not correspond in a level 8 etc. from the beginning, it is not related. The portion returned to a screen also rewrites a calculation value by C++ to C, and it is VectorC. It changes so that it may compile. The conversion to int from a float value is even slight. It should become quick and this should also contribute to some speed rises.

2003.02.20: Ver0.1.1 :
mode1 addition.
Already [ the name of WaveletReducer to ] is derailment feeling. っていう or って taken out the functional addition version within the day is how -- a thing not the knowledge that depends mode1 from a book for the time being but mounting from a flash it is -- it will be satisfied only with things It is the method by which it was all known in fact. Mind [ like ] has not been carried out, either.

2003.02.20: Ver0.0.1 : Reference-mounting. There is neither what 変哲 nor 面白み nor the feature.

2003.02.09: Ver0.0.0 :
Technical investigation start of wavelet deployment.
It creates by the name of wavelet_test.
It is made to degenerate by hard threshold.

oledoe is offline   Reply With Quote
Old 1st December 2004, 13:06   #3  |  Link
Registered User
Join Date: Oct 2001
Posts: 15


It is the number of times of analysis of a transverse direction. Please set up in 0-5.

The same as the above of the direction of length.

The plain which carries out wavelet deployment is set up. 0: Luminosity

threshold 1-5:
It is a 閾 value over each level.
An outline will be ruined if it raises too much.
Although there is no maximum, probably, it should carry out to about 20 usually.

It is the kind of MotherWavelet.
0: Haar (a part for 2Block)
1: N=2, Daubechies of (a part for 4Block)
2: N=3, Daubechies of (a part for 6Block)
3: N=2, Coiflet of (a part for 6Block)
4: N=4, Daubechies of (a part for 8Block)
5: N=6, Daubechies (part for 12Block)

Although it becomes so heavy that a numerical value is large, how for it to
Although Haar is the lightest, since it is easy to come out of a block noise in Wavelet degeneration Probably, it will be better not to use.
It is equal to use also by Wavelet of Haar at the time of multiplex resolution analysis decomposition 2D NR.
Even if it raises a 閾 value so that a numerical value is large, an outline stops it square to be able to become easily.
The method of NR is chosen.
0: NR by Wavelet degeneration
1: Multiplex resolution analysis decomposition 2D NR
2: NR by Wavelet degeneration (SoftThreshold)
3: NR by Wavelet degeneration (distribution correspondence SoftThreshold)
4: In SoftThreshold of NR(rate 算?k退)

mode 2 and 3 by Wavelet degeneration, it sets up whether which usually makes a Threshold value effective to a field by 0-100. In mode4, it came to use it for the setup of the strength of the degree of degeneration. The maximum is not established or more by zero at mode 4:00. It writes below in detail.

It is a flag for debugging. Please use it by 0 usually.
Since a wavelet deployment value can be seen, it may be good for adjustment of a parameter.
The technique of NR changes with the value of mode.

It is NR filter using wavelet degeneration. Since it is only the wavelet degeneration which what 変哲 does not have, either, it is built only in the theory which is fundamentally existing. Since it is the technology currently used also for the picture compression by wavelet, if a 閾 value is raised too much, naturally the picture of wavelet degeneration will be ruined. Sakka who saw if it was only mere a few, although it was also ruined the picture's mere to have raised a few in fact -- it is the hardly known level and the pixel in which only a few is different from the circumference is screened Although it does not disappear unless the pixel which a difference understands clearly by appearance raises a 閾 value considerably, if a 閾 value is raised to there, probably the roughness of a picture will be frightful. This neighborhood serves as a future subject.

wavelet deployment performs multiplex resolution analysis and give 2D NR to a Wavelet deployment value on each of the level. Corners is cut in fact for improvement in the speed, and the dot currently referred to is limited towards the Wavelet deployment. (Refer only to the dot of the direction of length only for the dot of a transverse direction by reference and Wavelet deployment of the direction of length in Wavelet deployment of a transverse direction) For the reason, it is ruined. Although unlike the method using wavelet degeneration a picture is seldom sometimes ruined in a gradation portion since it is the flat and smooth-ization NR fundamentally, a noise may remain in an outline portion. Moreover, if a 閾 value is raised too much, it will fade. However, the absolute value of a 閾 value is not those[ with a problem ]-ま(ed) even if it makes [ more ] some from mode0. When you use Wavelet of Haar, please make it this mode.

Although it is NR filter with which this also used Wavelet degeneration, confirm the value of Threshold here to the whole screen. Wavelet degeneration makes a portion with little change flat and smooth by setting the Wavelet deployment value below a 閾 value (Threshold value) to 0 -- although it is a technique, in mode2, it is made the value which pulled the Threshold value from the absolute value also to a Wavelet deployment value with the absolute value beyond this Threshold value It specifies whether rate (it is the technique generally called Soft Thresholding) with the effect which makes the whole screen flat and smooth by this operation pulls what% of a Threshold value, when pulling a Threshold value from a Wavelet deployment value. If it is made 100, the Threshold value itself will be pulled and the whole screen will become quite soft touch. If it is made 0, the completely same result as mode0 will be brought. Although the NR effect is higher, since BOKA [ this mode ] that there is comparatively no forgiveness, be careful of it and set up Threshold.

The area where the contrast of a screen of this mode is strong is the mode in which leave and the weak area of contrast is NR(ed) preponderantly. Specifically, a Wavelet deployment value is received. 1/9 1/9 1/91/9 1/9 1/91/9 1/9 1/9 the procession of the form to say -- hanging -- the center of a Wavelet deployment value -- NR is given to the portion of BETA, leaving an edge portion by giving NR strongly, even if it becomes a value only with one point higher [ even if it takes a preponderant average and becomes a deployment value smaller than a 閾 value partially in a portion with many Wavelet deployment values, it is weak in NR, and ] than a 閾 value in a portion with few [ conversely ] Wavelet deployment values In addition, since how to give NR itself is a SoftThreshold system as well as mode2, if rate is set up to 100%, it will become the same as mode2 as a result. I think that rate may be usually used at 0%. Probably, at rate0%, it will be good to make it mode2 or to raise a rate value to take the noise which can never be taken. In addition, before NR(ing), in order to create a Wavelet deployment value distribution map, it is heavier than other modes.
oledoe is offline   Reply With Quote
Old 1st December 2004, 13:09   #4  |  Link
Registered User
Join Date: Oct 2001
Posts: 15
waveletreducer.txt (continued)

Although all the Wavelet deployment values below a 閾 value will be set to 0, in a Wavelet deployment value, inclination of an incoming signal is maintaining a fixed state, and 0 may eliminate the shake of a delicate vector completely, and it may stop leading to the next block continuously by any techniques other than mode1. (Although it becomes a flat and smooth side completely in Wavelet of Haar, since it is influenced by the surrounding Wavelet deployment value when the other Wavelet is being used, a possibility of becoming a perfect flat and smooth side becomes so low that the value of type is high) . Then, it brings close to 0 more, so that the original Wavelet deployment value is close to 0, it divides so that the original Wavelet deployment value is a value far from 0 and it may not be made to decrease not much from the original value, and it is the degeneration method using 算. It is only made hard to be for this mode not to erase a noise and conspicuous. However, it can be said that it is the mode which cannot break down an outline easily. Moreover, the strength of effectiveness can be adjusted with the value of rate. It is not effective at all by 0. It is as an original formula at 100%. Effectiveness becomes strong rapidly more than by it. It restricts to this mode and a maximum is not in the value of rate. Since a 閾 value is not necessarily raised, the influence to the portion below a 閾 value can be strengthened without affecting a portion with the Wavelet deployment value beyond a 閾 value. However, since a block noise may occur if it raises too much not much, please make it moderate.

Below common-matter In order to make the number of times of wavelet deployment unrestricted as a matter of fact, the resolution of an input picture is once changed into the multiplier of 2. Since wavelet deployment is carried out after being changed into 1024*512, if it is the picture of 320*240 and is the picture of 512*256 and 720*480, it is very heavy. Since weight changes considerably also by levelX, Y, and type, please use it in the range which can be equal to use as a matter of fact. Probably, it will be better to stop and solve, if it did not have confidence considerably in a CPU power that type NR(s) [ level ] the picture of 720*480 by 5*5 4. Although it is NR, only luminosity is NR(ed) fundamentally. A reason is easy and is because five more arguments will increase if you are going to make it a color also NR. If there are too many arguments, please write twice by 鬱陶しい's, changing [ to NR a color, either ] the value of func. However, in almost all cases, an appearance problem becomes that there is nothing only by NR of luminosity. Although it thinks that waveletReducer.dll and waveletReducer_Athlon.dll are enclosed, it is the thing which sets up a compile option by the object for Pentium, or the object for Athlon, and recompiled it, respectively. the direction which the direction which is not attached at all attaches with the object for Pentium series (it corresponds after PentiumIII) and Athlon becomes Athlon series (3D Now -- loading -- indispensable) It has divided into C++ sauce and C sauce, and VisualC++6.0 and the C section are compiled for the C++ part by VectorC 1.3.3SpecialEdition (since there is no assembler ?曹_ capability, it depends on VectorC).

Parameter consideration: mode4 is the mode newly established for the purpose of ?A which does not break down an outline. Since it happened to have thought while groping for the method of NR(ing) without breaking down an outline, it mounted. However, since the effect of NR is low, it may be almost meaningless after all. Moreover, if a 閾 value is raised in order to raise an effect, although it is MASHI, an outline will be ruined from other modes, too. It says or fades. On the characteristic of Wavelet, although it thinks that it cannot fade easily rather than other NR(s) even if it raises a 閾 value earnestly, since it is hard to carry out error diffusion, when still fading, it fades. It says that it fades, or is melted and lost. Even if mode3 seldom raises a 閾 value, it is the mode newly established in order to make a noise easy to take. In raising a 閾 value to a barricade barricade earnestly, there is almost no merit. Merely, since it is only heavy, in such a case, please use mode 0, 1, and 2 obediently. However, although I think that it can NR effectively, how is it by animation from which the outline portion and the portion of BETA have parted clearly? Probably, it will be better to depend on mode2 to animation which it is too old, is full of the noises in a screen, and a noise and an effective signal call half-and-half. In mode3, I think that a noise is probably misunderstood with edge and a miserable result is brought. Moreover, isn't it good like a DVD lip to sauce in which a noise is hardly contained primarily at mode1? In mode3, since the low portion of contrast disappears after all, delicate nuance could not remain easily and I think whether there is any problem which was alike also by mode0. Although it may be hard to use mode 0 and 3 properly, the direction of mode3 thinks a little that the NR effect becomes high. However, as written also in the top, since mode3 is the system which deletes the low portion of a contrast difference, leaving the intense portion of a contrast difference, it thinks that the direction of mode0 is suitable to a scene, an on-the-spot photo, etc. to which the whole screen faded. Probably, it will be better to seldom use, since a block noise tends [ quite ] to come out of type0 except mode1 (calculation is high speed). If it is a type value beyond it, although there is no so big difference, it will change [ that how for it to be / the circumference's of an outline / ruined is various and ]. If a type value is large, although roughness will tend to spread comparatively broadly, the roughness itself instead becomes thin. Although roughness will stop at the circumference of an outline if a type value is low, the contrast of roughness becomes high, and even if it is going to take by 2D NR later, it may suffer troubles.

Future view etc. : * Don't obscure an outline, giving NR effectively about the noise reduction effect. Although it is the mission imposed on NR program, it is a pattern still far from the fresh ground. Still, this program in this time is not practical. But there is [ most ] no case where WaveletNR is the most practical. (Only usual 2D NR will be sufficient for business in general.) It is necessary to carry out to Wavelet truly. It is a rare case. However, it will be that it is not impossible, since a great senior called 某 and Type-G exists. Although the present target of waveletReducer is called Avisynth version Type-G, it is in the state which is going to grope for many things and is going to progress in front little by little. Since a certain amount of indicator for not obscuring an outline for the time being has stood, it will heighten the NR effect more, the following target maintaining this outline. It is groping for the way of fusion to usual 2D NR experimentally for the time being. Although it is a stage still far from practical use, I think that the advantage of skillful _使えば Wavelet and the advantage of 2D NR can be united. Of course, 3DNR-ization is considered the back. Although it thinks that it becomes mounting with another function, probably, the formation of 3DNR does not have mounting, since it is the good method of increasing the number of effective samples by mistake. Incidentally the relation of two points which adjoin each other on a level 1 is summarized to Wavelet-ization, and the two points are summarized to one point. On a level 2, two points collected on the level 1 are summarized to two-point Wavelet-ization, and it is summarized to one point. (That is, it is set to NR of the range of 4*4) It is the same as the above about two points settled on the level 2 in the level 3. since -- it continues with 16*16 and 32*32 the range of 8*8, and henceforth Since it is such a system, in a scene from which the color is changing every other dot, for example, it is possible to take whole color MURA, without breaking down change of the color in every dot. (threshold of a level 1 is pressed down lowness in this case, and a level 2 and threshold of 3 are raised) It may be difficult to take a merely still really severe noise something. Wavelet extracts the high frequency ingredient more than fixed of the picture currently operated as a feature point, in Wavelet degeneration, is deleting the portion (that is, a Wavelet deployment value's being low) which is not still so high as for frequency among the extracted feature point, and cuts down MURA, with the portion of the feature left. In the multiplex resolution analysis decomposition NR, when a value is less than a fixed range in two points which adjoin each other among Wavelet deployment values, by making the portion flat and smooth, it turns flat and smooth and the delicate singular point generated in a gently-sloping change is assimilated to a surrounding change (that is, when of the same grade in frequency change). Since it is premised on a noise being not much conspicuous by neither of the cases, the conspicuous noise is conversely extracted as a feature and cannot be NR(ed). If a 閾 value is raised so that the conspicuous portion can be NR(ed), it will delete to other significant portions. It is mode2 that it is going to take a noise even if it deletes the information on a significant portion. In mode2, since Threshold*rate% of value is subtracted also to the Wavelet deployment value beyond a Threshold value, the part effective signal ingredient will decrease. However, the noise which cannot be taken can also be taken in mode 0 and 1. In addition, if rate is 0, it will be called mode2=mode0. Moreover, a Threshold value = as for mode 0 and 2, also in infinity, it becomes equivalent as a result (it is a talk in theory, and on the fact, when a Threshold value is 1280 or more, mode 0 and 2 becomes equivalent).

Some proposals of improvement in the speed are also proposed, and they are due to be mounted one by one. Although it is a translation [ parameter / other NR(s) with 弄_り ] in AVIUtil, still more, far one is visible to practical use, and it becomes it that it is sometimes likely to be discouraged. This program is still the thing which is done and between does not have, either. Please watch by the lukewarm eye. Notes: Even if you cover to some patents and you make [ it is sufficient and ] it them, please laugh and allow. Though it suffers a certain damage using this program, I need your help by accountability. Bibliography : Joint establishment publication The signal processing and the image-processing Asakura bookstore by wave let Wavelet transform and its application Softbank C Magazine February, 2003 issue

oledoe is offline   Reply With Quote
Old 1st December 2004, 17:44   #5  |  Link
Mug Funky
interlace this!
Mug Funky's Avatar
Join Date: Jun 2003
Location: i'm in ur transfers, addin noise
Posts: 4,555
or you could try VagueDenoiser
sucking the life out of your videos since 2004
Mug Funky is offline   Reply With Quote
Old 1st December 2004, 23:17   #6  |  Link
Registered User
Join Date: Oct 2001
Posts: 15
Vaguedenoiser doesn't work for me

Thanks for the tip.
I tried the last version of vaguedenoiser, and it doesn't work in my PC. I have a Celeron 400 (without SSE). Maybe that's the reason.
oledoe is offline   Reply With Quote
Old 2nd December 2004, 17:37   #7  |  Link
AviSynth plugger
Fizick's Avatar
Join Date: Nov 2003
Location: Russia
Posts: 2,183
Thanks for info about WaveletReducer.

But it is strange, that Vaguedenoiser not works. It must works in C code for old CPU. I will check code, but i have not Celeron 400 now.
My Avisynth plugins are now at http://avisynth.org.ru and mirror at http://avisynth.nl/users/fizick
I usually do not provide a technical support in private messages.
Fizick is offline   Reply With Quote
Old 3rd December 2004, 11:56   #8  |  Link
Registered User
Join Date: Oct 2001
Posts: 15

I posted this in another thread by mistake.
I post it here again, so that everything is in one place.
I can't delete the other thread, though. It seems administrative rights are needed for that. Sorry.




level: It is the number of times of analysis.
Please set up in 0-4.

func: The plain which carries out wavelet deployment is set up.
0: Luminosity

threshold 1-5:
It is a 閾 value over each level.
An outline will become thin if it raises too much. It fades and is spread around. Although it is hardly, since it becomes the textures carried out smoothly, let's make it moderate. There is no maximum.

It is the kind of MotherWavelet.
0: Haar (a part for 2Block)
1: N=2, Daubechies of (a part for 4Block)
2: N=3, Daubechies of (a part for 6Block)
3: N=2, Coiflet of (a part for 6Block)
4: N=4, Daubechies of (a part for 8Block)
5: N=6, Daubechies (part for 12Block)
It becomes so heavy that the numerical value of is large. moreover, unlike NR using Wavelet degeneration, the technique of NR itself is traditional the view seen also by Wavelet of Haar for 2D NR -- it hardly becomes amusing However, a level 4 and the 閾 value of 5 neighborhoods A block noise can be seen, while using Haar, if it is made a very big value. In this case, type Please enlarge a value.

Description : It is 2D NR using the sub band decomposition by Wavelet conversion which can set up a 閾 value for every frequency. the side effects generated in WaveletReducer since it is not operated [ as opposed to / a Wavelet deployment value / especially ] -- all hardly occur However, since the NR technique is usual moving average type 2D NR, the energy which increased by the noise signal does not disappear. Although a noise ingredient melts into the surroundings, since the signal of the part has increased, it changes appearance with NR using Wavelet degeneration. But since a noise signal cannot necessarily be completely omitted by NR using the usual Wavelet degeneration, it may seldom change. Since it has the function in which effect can be adjusted for every frequency which is one of the features of NR using Wavelet, one to use Wavelet can be achieved now. The algorithm of used 2D NR performs a ??value check about 8 blocks the surroundings of an observing point first, and the absolute value of a difference with an observing point counts ten or more significant work among the things within a ?ン定閾 value. It is here. The absolute values of a difference with an observing point are ten or more things. Effective sample total within the 閾 value in /8 blocks When a の value is 2/3 or more, 1 1 11 0.3 1 / 8.31 1 1 The procession to say is used. Moreover, when the value like the point is 1/3 or less, it is. 1 1 11 2 1 / 101 1 1 The procession to say is used. In the cases of other than the above 1 1 11 1 1 / 91 1 1 It comes to use the procession to say. Here, although there is an absolute value of a difference with an observing point or more with ten, a basis is not in the number of this 10 yet. Adjustment of this value may be attained in the future. Or it may come to calculate the optimal value by calculation.

Future view etc. : It is one of fusion experiments with 2D NR and Wavelet. it is this NR 自体ま -- although it is a thing with the room of research, is it the grade of being as accelerate たり [, and ]? [ that about two additional parameters increase at most ] But since it thinks that side effects are NR of ?_な目 comparatively, as for the man that have too much [ a CPU power ] and it is unavoidable, using may also be good.
oledoe is offline   Reply With Quote
Old 3rd December 2004, 12:03   #9  |  Link
Registered User
Join Date: Oct 2001
Posts: 15
@Fizick: you're welcome

You're welcome.

I appreciate your interest in my problems with VagueDenoiser, but I have to upgrade my machine once and for all.
It's not fair to expect support from opensource authors when my setup is not the standard one.
That would be time better spent working on fft3d

Last edited by oledoe; 3rd December 2004 at 12:32.
oledoe is offline   Reply With Quote

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT +1. The time now is 22:00.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, vBulletin Solutions Inc.