CN102054279A - Generating method of random patterns - Google Patents

Abstract

The invention provides a generating method of random patterns, comprising the process of generating a second pattern by adopting a filter which at least eliminates or reduces a low spatial frequency component whose spatial frequency is less than a specific value from the spatial frequency component of a first pattern which is configured with a plurality of points at random or with a radiance distribution; generating a third pattern which is shifted into a discretization information by adopting a shaking tuning method to the second pattern; generating a fourth pattern by adopting a Monte Carlo method to move isolated pixels.

Description

The generation method of random pattern

Technical field

The present invention relates to the generation method of homogeneity and randomness excellent pattern.

Background technology

At present, in display device, particularly widely used LCD, use the device that carries out light scattering mostly.For example, being used for will be in the light scattering of the light guide plate at the back side that is positioned at LCD transmission in the device of front (dot pattern that Chinese white constituted that forms based on the rear side surface in light guide plate etc.), the device (diffuser plate and light diffusion film etc.) that the light that is positioned at the light source under the liquid crystal panel is spread equably, the device (prismatic lens and two sides convex lens etc.) that is used for making the direction of light unanimity equably and takes out, be positioned at the non-glare treated device of mirroring (anti-dazzle film etc.) of the distinctness of the most surperficial of display device and inhibition surrounding enviroment, and the diffusion reflector of reflection display device use etc.This light scattering or light disperser are usually by disposing concavo-convex the realization at substrate surface.

The concave-convex surface structure can carry out in the following way to the giving of component parts of aforesaid display device, promptly, generation becomes the pattern (hereinafter referred to as fundamental design) on the basis of this concave-convex surface structure pattern, based on this fundamental design, at substrate surface processing concaveconvex shape.At this moment, require fundamental design can think full and uniform by the people.This is because when having uneven part (so-called inequality) on the relief pattern of the component parts that is carrying out light scattering, this inhomogeneous part can be directly linked to the inequality of demonstration or the unevenness of lightness.

In addition, fundamental design generates with computing machine mostly, but the pattern that generates with computing machine has limited size usually, therefore by being that fundamental design forms under the situation of sag and swell with the pattern that a plurality of this patterns are repeated to be in juxtaposition, when the homogeneity of the pattern with finite size that is generated is insufficient, observes repetition decorative pattern etc. and will bring harmful effect the quality of display device.

With regard to fundamental design, can ask with homogeneity is at random.The meaning at random is meant does not find systematicness in sufficient distance.The example that non-sag and swell fully at random gives the influence of display device is the phenomenon that is called moire fringe.Display device has the structure of having arranged pixel regularly.In the time will having the tectonic stacking of the systematicness different slightly, will produce the light and shade inequality that is called moire fringe with the systematicness of this pixel.Learn to have at non-sag and swell fully at random under the situation in cycle of hundreds of μ m degree in addition, can watch the pattern of iridescence.This is to cause by interfering by the light that structure reflected of rule.The phenomenon that iridescence appears in the card of laser disc (trade (brand) name) is the representational example of this phenomenon.This phenomenon also can be brought harmful effect to display quality.

Like this, the component parts of display device is being given under the situation of sag and swell, its employed fundamental design need have homogeneity and randomness concurrently.But, at present, have the generation difficulty of the pattern of the homogeneity of density and randomness concurrently.The method that the relief pattern of the component parts of aforesaid scattered light forms based on the fundamental design that makes a distribution is a usual way.But, make under the situation of a little arranging with uniform density, can near with two dimension in the closest packing structure of point be that the close systematicness of triangular crystal lattice is arranged, lose randomness.In order to address this is that, the whole bag of tricks on probation.

For example, (Japan) spy opens the 2002-14211 communique and discloses following technology, that is, have in the reflecting plate of concaveconvex shape on the surface, by recess or protuberance being configured to spiral fashion, the homogeneity of the relief pattern that gets both and randomness.But, in the method, be that there are the following problems under the situation of fundamental design: have good property in succession between pattern and the formation base pattern difficulty that just becomes at the pattern that repeats to be in juxtaposition with pattern with finite size.The meaning of this problem is can observe the lattice-like decorative pattern in specified period.Therefore, the inadequate pattern of the connectivity between relief pattern, in the application of the display device of large tracts of land progress in recent years, range of application is limited to can not directly visual place.In addition, noticeable in order not make the decorative pattern that is produced, the situation that the sheet plate that needs to be dispersed with particle is arranged at display device also exists.

(Japan) spy opens the 2005-215641 communique and discloses following technology, promptly, in the manufacturing of diffusion reflector in the employed photomask that is made of shading or printing opacity pattern, the systematicness of the triangular crystal lattice by destroying excellent in uniformity generates and has given the shading or the printing opacity pattern of randomness.But, following contradiction is arranged in this method, that is, if pay attention to randomness, the homogeneity sacrifice of the then density part of origination point, so density can involve uneven generation; On the contrary, if pay attention to homogeneity, then can not get pattern fully at random.

Summary of the invention

The objective of the invention is to, also excellent pattern of a kind of excellent in uniformity and randomness is provided.

To achieve these goals, what present inventors repeated to study with keen determination found that, according to following method, can realize also excellent pattern of excellent in uniformity and randomness, this method is, after having generated first pattern that constitutes by image and view data etc., by this first pattern being used the wave filter of removing at least or reducing the low spatial frequency composition of the not enough particular value of spatial frequency, generate second pattern, and this second pattern is used the tuning method of shivering, generate the 3rd pattern of the information that is transformed to discretize.Find in addition, as above-mentioned wave filter, can preferably use Hi-pass filter or bandpass filter, in the described Hi-pass filter, remove or reduce in the contained spatial frequency composition of first pattern and fix the low spatial frequency composition that the low spatial frequency of limit value B ' constitutes by bit, and the spatial frequency composition that extraction is made of the above spatial frequency of lower limit B ' (below, this lower limit B ' is also referred to as spatial frequency range lower limit B '); In the described bandpass filter, remove or reduce in the contained spatial frequency composition of first pattern by bit and fix low spatial frequency composition that the low spatial frequency of limit value B constitutes, and the high spatial frequency composition that constitutes by the spatial frequency that surpasses specific higher limit T, and extract the spatial frequency composition that constitutes by spatial frequency (below, will also be called spatial frequency range lower limit B, spatial frequency range higher limit T) at the lower limit B and the higher limit T of this particular range at the particular range of this lower limit B～this higher limit T.The present invention be based on such opinion further in addition all research finish.

The invention provides a kind of generation method of random pattern, possess: to dispose a plurality of points randomly or dispose luminance brightness (lightness: lightness) first pattern of Fen Buing, the wave filter of the low spatial frequency composition of the not enough particular value of spatial frequency is removed or is reduced in application at least from the contained spatial frequency composition of first pattern, generate the operation of second pattern; By described second pattern is used the tuning method of shivering, generate the operation of the 3rd pattern of the information that is transformed to discretize.

Above-mentioned the 3rd pattern can be the pattern with various progression institute discretizes such as 16 grades, 64 grades, 128 grades, but be applied at the random pattern that generates by the present invention preferably be transformed to pattern under the situation of job operation of the processing of resist operation and the suitable batch processes such as processing that print process realizes by the information of secondary discretize.

The generation method of random pattern of the present invention preferably also possesses the operation that generates the 4th pattern, wherein, to being transformed to the 3rd pattern by the information of secondary discretize, by Monte Carlo method isolated black or white pixel is moved, and generates the 4th pattern.

As the tuning method of shivering, preferred use error diffusion method.One of the inventive method preferred embodiment in, make the error-diffusion method of mapping fault diffusion by the scope below 6 pixels more than 3 pixels of being applied in, generate the 3rd pattern.

As above-mentioned wave filter, can preferably use the Hi-pass filter of from the contained spatial frequency composition of first pattern, only removing or reducing the low spatial frequency composition of the not enough particular value of spatial frequency.This Hi-pass filter is preferably only to be removed from the contained spatial frequency composition of first pattern or reduces spatial frequency less than 0.01 μ m ^-1The Hi-pass filter of low spatial frequency composition.

In addition, as above-mentioned wave filter, the preferred bandpass filter of using, in the described bandpass filter, by from the contained spatial frequency composition of first pattern, removing or reducing the low spatial frequency composition of the not enough particular value of spatial frequency and remove or reduce the high spatial frequency composition that spatial frequency surpasses particular value, extract the spatial frequency composition of particular range.

In the operation of above-mentioned generation second pattern, the lower limit B of the spatial frequency of the spatial frequency composition of the above-mentioned particular range that extracts by the application of bandpass filter is preferably 0.01 μ m ^-1More than and higher limit T be preferably 1/ (D * 2) μ m ^-1Below.At this, D (μ m) is based on printing equipment that the 3rd or the 4th pattern prints or based on the resolution of the device of the 3rd or the 4th pattern processing concaveconvex shape.In addition, by the inverse of the promptly maximum cycle length 1/B of inverse of the lower limit B of the spatial frequency of the spatial frequency composition of above-mentioned particular range and higher limit T be minimum cycle length 1/T according to

BandWidth(％)＝100×(1/B-1/T)/(1/B+1/T)

Represented BandWidth preferably satisfies following formula:

15≤BandWidth(％)≤70。

All excellent pattern of a kind of homogeneity and randomness can be provided according to the present invention.By the random pattern that method of the present invention obtains, can be preferably as being used for the fundamental design that for example display devices such as anti-dazzle film, diffuser plate, light diffusing sheet, light guide plate are given relief pattern with component parts is used; Perhaps, use, thus, can form homogeneity and randomness all excellent relief pattern or printed patterns as the fundamental design that is used to give printed patterns.

Description of drawings

Fig. 1 is the enlarged drawing of the preferred example expression generation method that can be used for random pattern of the present invention, by disposing first pattern that many points are generated randomly;

Fig. 2 be the expression generation method that can be used for random pattern of the present invention, by the figure of a preferred example that has determined first pattern that deep or light raster image constitutes by random number;

Fig. 3 is the figure that the part of first pattern shown in Figure 2 is amplified expression;

Fig. 4 be by many points dispose randomly two-dimensional array that first pattern (random dot pattern) that generated obtains by fast Fourier transform (FFT) be transformed to the spatial frequency distribution that the spatial frequency zone obtained an example, be transformed to the figure that an example of the spatial frequency distribution that the spatial frequency zone obtained is compared with the two-dimensional array that obtains by first pattern that has determined deep or light raster image (random number raster image) to constitute by random number by FFT;

Fig. 5 represents that the two-dimensional array that obtains by first pattern shown in Figure 1 is transformed to the figure of the spatial frequency resulting two-dimensional space frequency distribution in zone by FFT;

Fig. 6 is the result of amplitude correc-tion has been carried out in expression to the spatial frequency distribution shown in the dotted line of Fig. 4 the figure of an example;

Fig. 7 is the figure of an example of the shape that sees through frequency band in the space frequency strip of representing to extract by the application of Hi-pass filter (seeing through frequency band);

Fig. 8 is the figure of another example of the shape that sees through the frequency band peak value in the space frequency strip of representing to extract by the application of Hi-pass filter (seeing through frequency band);

Fig. 9 is the figure of another example of the shape that sees through the frequency band peak value in the space frequency strip of representing to extract by the application of Hi-pass filter (seeing through frequency band);

Figure 10 is the figure of an example of the shape that sees through the frequency band peak value in the space frequency strip of representing to extract by the application of bandpass filter (seeing through frequency band);

Figure 11 is the figure of another example of the shape that sees through the frequency band peak value in the space frequency strip of representing to extract by the application of bandpass filter (seeing through frequency band);

Figure 12 is the figure of another example of the shape that sees through the frequency band peak value in the space frequency strip of representing to extract by the application of bandpass filter (seeing through frequency band);

Figure 13 is the figure of another example of the shape that sees through the frequency band peak value in the space frequency strip of representing to extract by the application of bandpass filter (seeing through frequency band);

Figure 14 is the figure of another example of the shape that sees through the frequency band peak value in the space frequency strip of representing to extract by the application of bandpass filter (seeing through frequency band);

Figure 15 is an example of the two-dimensional space frequency distribution after the bandpass filter has been used in expression to first pattern with spatial frequency distribution shown in Figure 5 figure;

Figure 16 is the value of expression BandWidth and the figure of the peaked relation of coefficient of autocorrelation;

Figure 17 is the figure of relation of the generation number of the isolated point after expression BandWidth value and the error-diffusion method binaryzation of being carried out;

Figure 18 is an example of second pattern that bandpass filter generated is used in expression to first a pattern shown in Figure 1 enlarged drawing;

Figure 19 is used for the figure that the weighting to the diffusion of the mapping fault of the error diffusion matrix learnt usually describes;

Figure 20 is by using based on Floyd; The local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of the matrix of Steinberg;

Figure 21 is by using based on Jarvis the local figure that amplifies expression of an example of resulting the 3rd pattern of error-diffusion method of the matrix of Judis and Nink;

Figure 22 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of Stucki matrix;

Figure 23 is by using the local figure that amplifies expression of an example based on resulting the 3rd pattern of error-diffusion method of Sierra 3Line matrix;

Figure 24 is by using the local figure that amplifies expression of an example based on resulting the 3rd pattern of error-diffusion method of Sierra 2Line matrix;

Figure 25 is by using the local figure that amplifies expression of an example based on resulting the 3rd pattern of error-diffusion method of Sierra Filter Line matrix;

Figure 26 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of Burks matrix;

Figure 27 is by using based on Stevenson; The local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of Arche matrix;

Figure 28 is the local figure that amplifies expression of second pattern that uses in the generation of the 3rd pattern shown in Figure 20～27;

Figure 29 is the figure that passing through shown in Figure 20～27 compared based on the spatial frequency distribution of the 3rd pattern of the spatial frequency distribution of the 3rd pattern of the error-diffusion method institute binaryzation of various matrixes and passing threshold method institute binaryzation;

The figure that the situation that Figure 30 is the generation number of the isolated point that will take place when generating the 3rd pattern by the application based on the error-diffusion method of generally well-known error diffusion matrix, generated with the passing threshold method compares;

Figure 31 is that expression diffusion length is the figure of an example of 1 error diffusion matrix;

Figure 32 is that expression diffusion length is the figure of an example of 2 error diffusion matrix;

Figure 33 is that expression diffusion length is the figure of an example of 3 error diffusion matrix;

Figure 34 is that expression diffusion length is the figure of an example of 4 error diffusion matrix;

Figure 35 is that expression diffusion length is the figure of an example of 5 error diffusion matrix;

Figure 36 is that expression diffusion length is the figure of an example of 6 error diffusion matrix;

Figure 37 is that expression diffusion length is the figure of an example of the error diffusion matrix of 3+4;

Figure 38 is that expression diffusion length is the figure of an example of the error diffusion matrix of 4+5;

Figure 39 is that expression diffusion length is the figure of an example of the error diffusion matrix of 3+4+5;

Figure 40 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 31;

Figure 41 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 32;

Figure 42 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 33;

Figure 43 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 34;

Figure 44 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 35;

Figure 45 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 36;

Figure 46 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 37;

Figure 47 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 38;

Figure 48 is by using based on the local figure that amplifies expression of an example of resulting the 3rd pattern of the error-diffusion method of matrix shown in Figure 39;

Figure 49 is with by the generation number of using the isolated point that takes place when error-diffusion method based on the error diffusion matrix shown in Figure 31～39 generates the 3rd pattern, the figure that compares with the situation of passing threshold method generation;

Figure 50 is by using the figure that compares based on the spatial frequency distribution of the pattern of the spatial frequency distribution of the 3rd pattern of Figure 40～48 of the error-diffusion method institute binaryzation of the error diffusion matrix shown in Figure 31～39 and passing threshold method institute binaryzation.

Figure 51 is the figure of example of the disposal route of the isolated point finished of expression Monte Carlo method;

Figure 52 is the figure of expression based on the variation of the 4th pattern of Monte Carlo method number of applications;

Figure 53 is the figure of relation of the generation number of expression Monte Carlo method number of applications and isolated point;

Figure 54 is the local figure that amplifies expression of random pattern (the 4th pattern) that embodiment 1 generates;

Figure 55 is the local figure that amplifies expression of pattern that comparative example 1 generates;

Figure 56 is the spatial frequency distribution of the random pattern shown in Figure 54, the figure that compares with the spatial frequency distribution of the pattern shown in Figure 55;

Figure 57 is the local figure that amplifies expression of pattern that comparative example 2 generates;

Figure 58 is the figure that the spatial frequency distribution of the spatial frequency distribution of the random pattern shown in Figure 54 and the pattern shown in Figure 57 is compared;

Figure 59 is the local figure that amplifies expression of random pattern (the 4th pattern) that embodiment 2 generates;

Figure 60 is the figure of the spatial frequency distribution of the random pattern that generates of expression embodiment 2;

Figure 61 is the local figure that amplifies expression of random pattern (the 4th pattern) that embodiment 3 generates;

Figure 62 is the figure of the spatial frequency distribution of the random pattern that generates of expression embodiment 3;

Figure 63 is the local figure that amplifies expression of random pattern (the 4th pattern) that embodiment 4 generates;

Figure 64 is the local figure that amplifies expression of pattern that comparative example 3 generates;

Figure 65 is the figure that the spatial frequency distribution of the spatial frequency distribution of the pattern shown in Figure 63 and the pattern shown in Figure 64 is compared;

Figure 66 is the local figure that amplifies expression of the employed first pattern A of the generation of pattern 1;

Figure 67 is the local figure that amplifies expression pattern 1;

Figure 68 is the local figure that amplifies expression pattern 2;

Figure 69 is the local figure that amplifies expression pattern 3;

Figure 70 is the local figure that amplifies expression of the employed first pattern B of the generation of pattern 4;

Figure 71 is the local figure that amplifies expression pattern 4;

Figure 72 is the local figure that amplifies expression pattern 5;

Figure 73 is the local figure that amplifies expression pattern 6;

Figure 74 is the local figure that amplifies expression of the employed first pattern C of the generation of pattern 7;

Figure 75 is the local figure that amplifies expression pattern 7;

Figure 76 is the local figure that amplifies expression pattern 8;

Figure 77 is the local figure that amplifies expression pattern 9;

Figure 78 is the local figure that amplifies expression of the employed first pattern D of the generation of pattern 10;

Figure 79 is the local figure that amplifies expression pattern 10;

Figure 80 is the local figure that amplifies expression pattern 11;

Figure 81 is the local figure that amplifies expression pattern 12;

Figure 82 is the local figure that amplifies expression of the employed first pattern E of the generation of pattern 13;

Figure 83 is the local figure that amplifies expression pattern 13;

Figure 84 is the local figure that amplifies expression pattern 14;

Figure 85 is the local figure that amplifies expression pattern 15;

Figure 86 is the figure of the spatial frequency distribution of expression first pattern A～E;

Figure 87 is the figure of the spatial frequency distribution of expression pattern 1～3;

Figure 88 is the figure of the spatial frequency distribution of expression pattern 4～6;

Figure 89 is the figure of the spatial frequency distribution of expression pattern 7～9;

Figure 90 is the figure of the spatial frequency distribution of expression pattern 10～12;

Figure 91 is the figure of the spatial frequency distribution of expression pattern 13～15;

Figure 92 is the conclusion figure of reduction degree of the low spatial frequency composition that difference realized of the method for making of pattern.

Figure 93 is the figure of relation of the generation number of expression method for making of pattern and isolated point;

Figure 94 is the local figure that amplifies expression of first pattern that disposes light brightness distribution randomly;

Figure 95 carries out the binaryzation of the application of Hi-pass filter and threshold method and the local figure that amplifies expression of pattern that obtains to first pattern shown in Figure 94;

Figure 96 carries out the binaryzation of the application of Hi-pass filter and error-diffusion method and the local figure that amplifies expression of pattern that obtains to first pattern shown in Figure 94;

Figure 97 carries out the application of the binaryzation of application, error-diffusion method of Hi-pass filter and Monte Carlo method and the local figure that amplifies expression of pattern that obtains to first pattern shown in Figure 94;

Figure 98 is the figure of the isolated point generation number of the pattern shown in expression Figure 95～97;

Figure 99 is the figure that the spatial frequency distribution with the pattern shown in Figure 94～97 compares;

Embodiment

The generation method of＜random pattern 〉

Below, preferred forms of the present invention is elaborated.Being characterized as of the generation method of random pattern of the present invention, for example, after having generated first pattern that constitutes by the pattern that disposes many points randomly or pattern of disposing light brightness distribution etc., first pattern used from the contained spatial frequency composition of first pattern, remove at least or reduce wave filters such as the Hi-pass filter of low spatial frequency composition of the not enough particular value of spatial frequency or bandpass filter, generate second pattern, resulting second pattern is transformed to the information of discretize by the tuning method of shivering, generates the 3rd pattern.In addition, as described later, the isolated point that the 3rd pattern is contained is handled by Monte Carlo method, is made as the 4th pattern and also can.

At least remove or reduce the wave filter of low spatial frequency composition of the not enough particular value of spatial frequency and the method for the present invention of the tuning method of shivering according to having used, can obtain the pattern that homogeneity and randomness have concurrently, therefore, by this pattern is used as fundamental design, can form the relief pattern or the printed patterns of homogeneity and randomness excellence.At this, in the present invention, so-called pattern is that the meaning of " evenly (having homogeneity) " is not must be ground, space state fully uniformly, is meant no matter implemented pattern formation, and the people thinks to have inhomogeneity state.Human eye does not have can be familiar with this specific character singly with tiny to a certain degree variation, therefore, and to stipulate that pattern quilt understanding in human eye that above tiny degree changes is uniform pattern.This meaning is that the low spatial frequency composition in the spatial frequency composition that has of pattern is few.Therefore, in other words " the evenly pattern of (having homogeneity) " can be the few pattern of low spatial frequency composition.In the present invention, " homogeneity " of aforesaid pattern is to realize by using the wave filter remove at least or to reduce the low spatial frequency composition of the not enough particular value of spatial frequency.

According to the present invention, therefore can obtain the few pattern of aforesaid low spatial frequency composition, also can make by the wad a quilt with cotton phenomenon of unrest of the show uniformity that is called " flicker " that to be relief pattern that fundamental design was processed into this pattern produce with the combination of the color filter of display device and be suppressed.Promptly, according to the present invention, can generate and the few pattern of the corresponding low spatial frequency composition of the Pixel Dimensions of color filter, therefore, can form and have and same degree of the Pixel Dimensions of color filter or the relief pattern that lacks than the concaveconvex shape of its big slightly size (cycle), therefore can suppress " flicker " efficiently.

In addition, in the present invention, the meaning of pattern " being (having randomness) at random " be specialized range has been carried out when research, and near it similarity of existing pattern little.This similarity can be estimated by autocorrelation analysis.Auto-correlation is the method that self inner existing homophylic intensity is quantized.

The method according to this invention, owing to be to have used to remove at least or reduce the wave filter of low spatial frequency composition of the not enough particular value of spatial frequency and the tuning method of shivering, further use the method for Monte Carlo method as required, therefore can obtain with existing method random pattern that can not realize, that suppressed the low spatial frequency composition.This random pattern that has suppressed the low spatial frequency composition is: the pattern after the inequality of being recognized by human eye is able to highly suppress.In addition, because the pattern that can realize making period of change to suppress to heavens, therefore to make the fundamental design of anti-dazzle film that flicker suppressed etc. be splendid as for example being used to make.In addition, the method according to this invention owing to can obtain the little random pattern of systematicness, is the display device component parts that fundamental design has been implemented concavo-convex processing with this pattern therefore, with the combination of pixels situation together of color filter under, difficult generation is Ah's interference not.In addition, as removing at least or reducing under the situation of filter applies bandpass filter of low spatial frequency composition of the not enough particular value of spatial frequency, the high spatial frequency composition of concavo-convex processing and printing processing difficulties is suppressed, and therefore can improve the machining reproducibility of relief pattern and printed patterns.

In addition, the connectivity between the pattern of the random pattern that is obtained by method of the present invention when repeating configuration side by side is good, and being applied to concaveconvex shape when giving of display device component parts waited, is difficult for the disadvantageous periodic lattice-like decorative pattern of generation.

At this, the meaning of " pattern " of " first～the 4th pattern " be image, view data, discretize information two-dimensional array or be disposed at the array of the opening of plate.

Above-mentioned view data also can be the view data (raster image) of raster mode, also can be the view data (vector image) of vector form.Raster image is the data that image is showed as the enumerating of point (point) of being with look.In raster image, the look information of each point is preserved with numerical value.As the form of preserving this raster image, exist variously, but, enumerate for example bitmap as common especially form.As bitmap, widely usedly especially be: 24 colored bitmaps that the intensity of red, green, blue is represented by 8 bit depth respectively, luminance brightness are by 8 bit depth, 256 grade (Duan Bands) 8 gray bitmaps of expression.

As the form of preserving raster image, except that bitmap, can enumerate the view data of having used compression algorithm etc. is PNG (Portable Network Graphics), TIFF (Tagged Image FileFormat), JPEG, GIF various forms such as (Graphics Interchange Format).

In vector image, if the coordinate of the terminus of line (position), curve, then its crooked mode, fineness degree, look, can be preserved with numerical value by the information such as look of the face that these lines surrounded.The image that records the set of these numeric datas or radius of a circle and centre coordinate, polygonal each punctuate coordinate etc. also is contained in vector image.

As the form of preserving vector image, as common especially form, illustration DXF (DrawingInterchange File), SVG (Scalable Vector Graphics).But in the present invention, vector image is not limited to these illustrative forms so long as belong to the image of above-mentioned definition and get final product.In addition, vector image is not limited to two dimension, also can be the image with three-dimensional information.

In addition, have the circle of sealing and the image of polygonal array in the vector image and can easily replace with above-mentioned " in the array of the opening that plate disposed ".

As mentioned above, pattern of the present invention is not limited to as image or the handled pattern of view data, also can be the pattern of giving as the two-dimensional array of the information of discretize.As the method for the information of preserving discretize, can enumerate floating-point (for example, 64 floating-points), integer various forms such as (for example, signed 32 integers, not signed 16 integers).

(generation of first pattern)

As first pattern, can use pattern optional from the pattern of above-mentioned definition, can be pattern arbitrarily with deep or light or numerical value change.More specifically, can enumerate the two-dimensional array etc. of the information of pattern that the view data (disposing view data a plurality of white points or dispose a plurality of stains on white background etc. on the black matrix) that for example disposes a plurality of points in the gamut of image, the pattern with deep or light variation etc. have light brightness distribution, discretize, in addition, when first pattern being used wave filter such as Hi-pass filter or bandpass filter (about this point, the back is narrated), carry out under the situation of Fourier transform with optical means, also can be the plate that disposes opening.In addition, forming the pattern that is attached with toner on figuratum dried photocopy (dry plate) and the transparent base partly, also can be used as first pattern and use.The configuration of the opening in configuration, light brightness distribution and the plate of the point in the view data etc., both can be systematicness, also (irregular) at random, but in the spatial frequency zone can obtain not only on a large scale, have amplitude and the pattern of systematicness low (randomness height), therefore preferably be made as configuration at random.

Describing many points in the gamut of the image that generates passing through randomly generates under the situation of first pattern, as the device of describing many points randomly, can enumerate: for example the image of width W X, height WY is generated the simulation random number series R[n take 0～1 value], the x coordinate of dot center is WX * R[2 * m-1 thereby for example generate], the y coordinate is WY * R[2 * m] the method for many points.At this, n, m are natural numbers.As the method that generates the simulation random number series, so long as the randomizer subtraction algorithm of linear congruential method, Knuth, Xorshif or MT anolog random number generator etc. have can with the method for the corresponding sufficient Cycle Length of counting that is distributed, can use and simulate the random number method of formation arbitrarily.Perhaps, be not limited to simulate random number, also can utilize hardware, generate and arrange first pattern a bit randomly by generation random numbers such as thermal noises.

The shape of point can be that circle, ellipse etc. are justified shape and polygon etc., also can dispose many points with same shape, also can dispose the point of many different two or more shapes.In addition, the size of point both can be that all point is all identical, also can be different.Therefore, under the situation of point, also can generate first pattern, many points with multiple footpath are disposed randomly by many points with a kind of footpath (diameter of point) are disposed randomly for the circle shape.

The equalization point footpath (mean value in the point of the being had a few footpath in the pattern) that constitutes the point of first pattern is not particularly limited, but under the situation of using bandpass filter, be preferably set at the peak value that has a footpath in the scope by frequency band and do not have peak value being lower than this low spatial frequency zone by the scope of frequency band, therefore be generally 4～50 μ m, be preferably 16～32 μ m.Directly surpass under the situation of 50 μ m at equalization point, have the tendency that comprises the low spatial frequency composition in resulting second pattern morely, on second pattern, easily produce gradation unequal.On the other hand, too small in the equalization point footpath of the point that constitutes first pattern and use bandpass filter time marquis, under the less situation of the amplitude of the spatial frequency composition that is extracted, the randomness that first pattern has is vulnerable to infringement, can not obtain preferred second pattern.Equalization point footpath is preferred uses and gives the spatial frequency range higher limit T of bandpass filter and greater than 0.5 * (1/2 * T)).Thus, under the situation of filling rate in preferred range described later of point, easily generation fully comprises spatial frequency composition that is extracted by bandpass filter and second pattern that is difficult for producing gradation unequal.

The situation of use Hi-pass filter similarly, be preferably set at the peak value that has a footpath in the scope by frequency band and do not have peak value being lower than this low spatial frequency zone by the scope of frequency band, therefore the equalization point that constitutes the point of first pattern directly is generally 4～50 μ m, be preferably more than the 6 μ m, more preferably more than the 8 μ m, in addition, be preferably below the 32 μ m, more preferably below the 30 μ m, more preferably below the 12 μ m.Directly surpass under the situation of 50 μ m at equalization point, comprise the low spatial frequency composition in resulting second pattern morely, easily produce gradation unequal on second pattern.

Filling rate (occupied area of the point in the image entire area) at the point when disposing many points and generate first pattern is preferably 20～80%, more preferably 20～70%, more preferably 30～70%, further be preferably 30～60%, be preferably 40～60% (for example, also can be about 50%) especially.Under the situation of the filling rate less than 20% of the point of count few and first pattern, can produce the inequality that the distinctive pattern by concentric circles constitutes on second pattern that is generated, the tendency that can not obtain preferred random pattern exists.In addition, surpass under 80% the situation too, have the tendency that inequality that the circular pattern by sealing constitutes can present morely, diminish randomness at the filling rate of point.

The view data that first pattern both can be made as vector form generates, and the view data that also can be made as raster mode generates.Under the situation of raster mode, can with 1,2,8 etc. arbitrarily the image format of bit depth generate first pattern.When the view data that is made as raster mode generates first pattern, preferably generate with higher sharpness in the detailed mode that can describe pattern.Preferred sharpness is for for example more than the 6400dpi, more preferably more than the 12800dpi.

Fig. 1 is the enlarged drawing of the preferred example expression generation method that can be used for random pattern of the present invention, by disposing first pattern that many points are generated randomly.First pattern shown in Figure 1 is the gray level image of 8 tones, and the black circular regions is a point 1.In the present invention, the diameter of putting is made as " some footpath ", the mean value in the footpath of the point of being had a few in the pattern is made as " equalization point footpath ".The equalization point of first pattern shown in Figure 1 directly is 16 μ m.In addition, image definition is 12800dpi.That is, the size of 1 pixel is equivalent to 2 μ m in length and breadth.In first pattern shown in Figure 1, size of images is WX=0.512mm, WY=0.512mm, and the filling rate of point is about 50%.In addition, the simulation random number of the centre coordinate of commit point is by mainly giving numerical value 607 and generate SIMDoriented Fast Mersenne Twister program, the SFMT ver1.3.3 that is installed by the research group of Hiroshima University.

In addition, as first pattern, also preferred use is disposed the pattern of light brightness distribution, has for example been determined deep or light raster image by random number.The concentration of each pixel (pixel) by raster image is decided by random number or the simulation random number that generated by computing machine, can obtain the little pattern of systematicness.

About the determining method of pixel concentration, describe as example with the situation of the simulation random number of the real number that uses output 0～1 scope.The tone number of pixel can be any number, but the easy to handle depth of shade is 1,8,16,24 etc., is preferably 8 (256 tones: index 0～255).For example, under the situation of 8 tones, by to having the PIXCEL[x of 8 bit depth, y] substitution PIXCEL[x, y]=R[x+y * Imagewidth] * 255, image can be generated.At this, x, y are the pixel coordinate of image, and Imagewidth is the image of x coordinate) width.In this embodiment, generate average index and be 127～128 image, but, also can generate the different image of mean value by additional offset.

Fig. 2 is the figure of expression by an example that has determined first pattern that deep or light raster image constitutes by random number, and Fig. 3 is its local figure that amplifies expression.Raster image shown in Figure 2 is the image by 8 tones that the luminance brightness of 1 pixel, 1 pixel is generated by the decision of simulation random number, particularly, by to having the two-dimensional array PIXCEL[x of 8 bit depth, y] substitution PIXCEL[x, y]=R[x+y * Imagewidth] * 255 and generate.At this, x, y are the pixel coordinate of image, and Imagewidth is the pixel wide of x coordinate.As array R[], the simulation random number series that the randomizer subtraction algorithm of the Knuth that takes the value between 0.0 and 1.0 that " NET Framework2.0 class method storehouse " contained Random class NextDouble method that use is developed by MicrosoftCorporation is generated draws.

In addition, first pattern also can be the two-dimensional array with above-mentioned raster image information that similarly generate, discretize.In this case, in order to determine each key element value of array, use the simulation random number.

The form of first pattern can be by for example being used to use Hi-pass filter or bandpass filter method, and with random pattern of the present invention process the device of concaveconvex shape as fundamental design or the form of the desired input of printing equipment of printing etc. suitably selected, especially, determined deep or light raster image owing to have amplitude by random number, therefore can preferably adopt in the wide spatial frequency range of width.This is because with irrelevant by the spatial frequency range of wave filters such as Hi-pass filter or bandpass filter extraction, easily keep the randomness of first pattern.

Fig. 4 be by many points dispose randomly two-dimensional array that first pattern (random dot pattern) that generated obtains by fast Fourier transform (FFT) be transformed to the spatial frequency distribution that the spatial frequency zone obtained an example, be transformed to the figure that an example of the spatial frequency distribution that the spatial frequency zone obtained is compared with the two-dimensional array that obtains by first pattern that has determined deep or light raster image (random number raster image) to constitute by random number by FFT, and be to be illustrated in spatial frequency 0～0.30 μ m ^-1The zone in the figure of intensity of amplitude, as shown in Figure 4, compare with the random number raster image, random dot pattern is particularly at 0.00～0.10 μ m ^-1The zone have higher oscillator intensity.In addition, about Fig. 4, be described later.

(generation of second pattern)

In the generation method of random pattern of the present invention, second pattern is to generate by first pattern is used the wave filter of removing at least or reducing the low spatial frequency composition of the not enough particular value of spatial frequency from the contained spatial frequency composition of first pattern.In the present invention, as this wave filter, can preferably use the Hi-pass filter of from the contained spatial frequency composition of first pattern, only removing or reducing the low spatial frequency composition of the not enough particular value of spatial frequency, perhaps preferred the use by from the contained spatial frequency composition of first pattern, removing or reducing the low spatial frequency composition of the not enough particular value of spatial frequency and remove or reduce the bandpass filter that high spatial frequency that spatial frequency surpasses particular value becomes to assign to extract the spatial frequency composition of particular range.Usually, pattern comprises and the corresponding spatial frequency composition of its variation.Change violent or dispose intensive pattern and comprise the high composition of many spatial frequencys, change slowly or composition that the spatial frequency of the pattern that configuration is evacuated is high less.By using Hi-pass filter or bandpass filter, can from the contained spatial frequency composition of first pattern, remove or reduce the spatial frequency composition of particular range, just the long period composition is the low spatial frequency composition.By using Hi-pass filter or bandpass filter, can reduce the low spatial frequency composition of second pattern, the 3rd pattern or the 4th pattern.Hi-pass filter or bandpass filter can be implemented by the sequence of operations of following (1)～(3) particularly to the generation of second pattern that the application of first pattern is finished.

(1) to the conversion in spatial frequency zone

At first, in order from the contained spatial frequency composition of first pattern, to extract specific spatial frequency composition (promptly, remove or reduce specific spatial frequency composition), when first pattern is raster image, as required, with first pattern be transformed to substitution have each pixel luminance brightness the floating type two-dimensional array g[x of corresponding value, y].At this, the position on the rectangular coordinate in x, the y representation raster image.By the two-dimensional array g[x to obtaining like this, y] be used for obtaining the device of size of the various spatial frequency composition of first pattern, can obtain representing spatial frequency composition that first pattern is contained and in the spatial frequency distribution of the amplitude of each spatial frequency.Device as the size that is used to obtain spatial frequency composition has optical means, mathematical method, and the method for obtaining is extensively generally adopted with particularly utilizing computer mathematics.The mathematical method that will obtain the size of spatial frequency composition usually is called Fourier transform.Fourier transform can be by utilizing computing machine discrete Fourier transformation (below, DFT) carry out.Therefore, the conversion to the spatial frequency zone can be by the two-dimensional array that is obtained by first pattern for example being utilized computing machine and using two-dimentional DFT and carry out.

As the DFT algorithm, can use generally well-known algorithm, particularly, therefore Cooley-Tukey type algorithm can preferably adopt owing to the computing velocity excellence.The DFT of Cooley-Tukey type algorithm be also referred to as fast Fourier transform (below, FFT).

Under the situation that first pattern generates with raster mode, the view data of this raster mode can easily be transformed to the spatial frequency zone on computers by using above-mentioned DFT algorithm.Generating and utilize above-mentioned DFT algorithmic transformation with vector form at first pattern is under the situation in spatial frequency zone, the view data of vector form is transformed to raster mode, the view data that will be transformed to raster mode then is transformed to two-dimensional array g[x, y on computers].At this, the position on the rectangular coordinate in x, the y representation raster image.Generating under the situation of first pattern as the common gray level image that for example has 8 tones, distributing 255, distributing 0 to black region to white portion.Utilize these values,, view data is transformed on computers the two-dimensional array G[f in spatial frequency zone by DFT _x, f _y].At this, f _x, f _yRepresent the spatial frequency of x direction, the spatial frequency of y direction respectively.In addition, under the situation that first pattern is endowed as the two-dimensional array of the information of discretize,, can be transformed to the two-dimensional array G[f in spatial frequency zone on computers by this first pattern is used DFT _x, f _y] be self-evident.

Utilizing under the situation of DFT, also can carry out from deduct the processing of whole key element average value P A of two-dimensional array as first pattern of the two-dimensional array of the information of discretize or each array key element of being transformed to first pattern of two-dimensional array.For example, can after will being transformed to two-dimensional array, carry out deducting the processing of whole key element average value P A of two-dimensional array from each array key element as first pattern that gray level image generated of 8 tones with 0～255 value.For example, be transformed under the situation of two-dimensional array, can obtain having the spatial frequency vector of amplitude sometimes in spatial frequency 0 at the gray level image of 8 tones that will have 0～255 value.This is caused to positive deviation by all key elements that constitute two-dimensional array.In this case, according in spatial frequency 0 and the processing that amplitude is 0 mode carries out deducting from each array key element whole key element average value P A of two-dimensional array also can.

Fig. 5 represents that the two-dimensional array that obtains by first pattern shown in Figure 1 is transformed to the figure of the spatial frequency resulting two-dimensional space frequency distribution in zone by FFT.In Fig. 5, the transverse axis and the longitudinal axis be the representation space frequency all.The point that diaxon is reported to the leadship after accomplishing a task is the point of spatial frequency 0, and far away more apart from this crossing (zero point), spatial frequency is big more.In addition, the intensity of the amplitude of each spatial frequency is represented by the concentration of look, means that painted dense more amplitude is big more.

As mentioned above, to be transformed to the spatial frequency zone resulting by FFT be as shown in Figure 5 two-dimensional signal to the image by 2-D data.But the demonstration of two dimension is not easy to understand thoroughly, therefore, below under the situation of representation space frequency distribution, can represent with the spatial frequency to be transverse axis and to be the spatial frequency distribution of the one dimension of the longitudinal axis with mean value in the oscillator intensity of each spatial frequency.With the figure that two-dimensional space frequency distribution shown in Figure 5 is represented by the spatial frequency distribution of one dimension, be the dashed curve figure of above-mentioned Fig. 4.That is, the dashed curve figure of Fig. 4 is expression will be transformed to the one dimension in spatial frequency zone resultant (result who is decomposed into spatial frequency by FFT is resultant) by the two-dimensional array that first pattern obtains by FFT the figure of spatial frequency distribution.In Fig. 4, transverse axis representation space frequency, the longitudinal axis is represented the mean value of the oscillator intensity of the key element under each spatial frequency.At this, oscillator intensity is meant the absolute value of each key element of two-dimensional array | G[f _x, f _y] |.In addition, with regard to mean value, in the time will being made as fmax, be divided into 128 parts and the key element of the two-dimensional array under each spatial frequency range of cutting apart averaged obtain by scope with spatial frequency 0～fmax by the resulting high spatial frequency of FFT.Spatial frequency range under the key element can be passed through by f _xAnd f _yThe value f that is calculated _aJudge.Below two formulas represent f and f _aCalculating formula be formula (A) and formula (B).

f?max＝(f _xmax ²+f _ymax ²) ^1/2 (A)

f _a＝(f _x ²+f _y ²) ^1/2 (B)

In addition, f _xMax represents f _xMaximal value, f _yMax represents f _yMaximal value.

Curve map shown in the dotted line of Fig. 4, even generated under the situation of first pattern by abundant simulation random number at random, first pattern also has the peak value of amplitude sometimes in specific spatial frequency.Exist under the situation of this amplitude peak, can not be by the specified spatial frequency lower limit of Hi-pass filter described later or bandpass filter specified spatial frequency range higher limit and lower limit, second pattern that obtains having desired spatial frequency characteristic, therefore preferably equate or the amplitude of each key element of mode correction about equally according to the amplitude of each spatial frequency in specific spatial frequency range.

Fig. 6 is the result of amplitude correc-tion has been carried out in expression to the spatial frequency distribution shown in the dotted line of Fig. 4 the figure of an example.Spatial frequency distribution (identical with the dashed curve of Fig. 4) before the amplitude correc-tion dots, and the spatial frequency distribution after the amplitude correc-tion is represented with solid line.In spatial frequency distribution shown in Figure 6,, be 0～about 0.30 μ m in spatial frequency by revising ^-1The zone, the amplitude constant of each key element.Like this, constant by in the spatial frequency zone that is obtained by Hi-pass filter or bandpass filter extraction amplitude being made, second pattern that generates by application Hi-pass filter or bandpass filter just has the spatial frequency composition of the particular range of uniform amplitude.This helps the control by the pattern properties that application generated of Hi-pass filter or bandpass filter.In addition, particularly, the correction of above-mentioned amplitude is by utilizing revised complex amplitude absolute value C by following formula: α=C/|A _Org| real number α that is given and complex amplitude A _OrgCarry out that multiplying carries out.Wherein, | A _Org| not null value.Therefore, above-mentioned correction can be at | A _Org| realize in the scope for nonzero value.

(2) application of Hi-pass filter or bandpass filter

Then, the two-dimensional array in the spatial frequency zone that obtained by DFT is implemented operation corresponding to Hi-pass filter or bandpass filter.By this operation, and make the contained low spatial frequency composition of first pattern remove or reduce.

Hi-pass filter is also referred to as high territory by wave filter, Low-Cut Filter, in the signal Processing field, has removing or reduces the effect of the composition of not enough appointed frequency.Corresponding to the operation of Hi-pass filter is operation in the spatial frequency composition that first pattern is contained, that removed or reduced and will be extracted by the spatial frequency composition that the above spatial frequency of this lower limit B ' constitutes by the low spatial frequency composition that constitutes than the low spatial frequency of spatial frequency range lower limit B '.About utilizing the situation of DFT, more specifically, be by array key element (real part of complex amplitude, each one of the imaginary part) substitution 0 (amplitude is made as 0) of the low spatial frequency composition of the scope of spatial frequency range lower limit B ' appointment or multiply by absolute value at the array that is transformed to the spatial frequency zone and contrast fully less than the operation of 1 value.As absolute value fully less than 1 value, when carrying out illustration according to the performance that is commonly referred to the wave filter of Hi-pass filter, enumerate absolute value for example from 0.5 near zero numerical value, absolute value from 0.3 near zero numerical value, absolute value from 0.1 near zero numerical value or absolute value from 0.01 near zero numerical value etc.Usually, the absolute value of multiplication value approaching more zero (comprising zero), desirable Hi-pass filter.

With regard to the value of spatial frequency range lower limit B ', relying in the spatial frequency that sees through ratio corresponding to Hi-pass filter is that can be considered as is the starting point of its rising edge under the rapid situation about rising in boundary as shown in Figure 7 with certain spatial frequency.On the other hand, seeing through under the mild situation about rising of ratio, the value of spatial frequency range lower limit B ' is set as the spatial frequency of 1/2 intensity that expression sees through the peak value of frequency band.About the spatial frequency range higher limit T of bandpass filter and spatial frequency range lower limit B too.Tranmittance rate shown in Fig. 7 and Fig. 8 described later～14 is represented the absolute value of the value that multiplies each other with above-mentioned each key element.In addition, in example shown below, all carry out operation corresponding to Hi-pass filter and bandpass filter by multiply by real number.

In the spatial frequency band that extracts by the application of Hi-pass filter (seeing through frequency band), each spatial frequency composition see through ratio (ratio of the oscillator intensity before the oscillator intensity after Hi-pass filter is used is used Hi-pass filter), as shown in Figure 7, seeing through the frequency band entire scope is constantly also can, as shown in Figure 8, its value variation also can.In addition, as shown in Figure 9, have a plurality of peak values through frequency band and also can.

Bandpass filter is also referred to as band filter, in the signal Processing field, have the intention of making scope frequency by and effect that the frequency beyond it is removed or reduced.Corresponding to the following operation of being operating as of bandpass filter, promptly, in the spatial frequency distribution of above-mentioned first pattern that obtains, remove or reduce in the contained spatial frequency composition of first pattern, and extract the spatial frequency composition that the spatial frequency by the particular range from this lower limit B to this higher limit T constitutes by the low spatial frequency composition that constitutes than the low spatial frequency of spatial frequency range lower limit B, and the high spatial frequency composition that constitutes by the spatial frequency that surpasses spatial frequency range higher limit T; Utilizing under the situation of DFT, more specifically, the operation of this bandpass filter is to not being included in by the array key element substitution 0 (amplitude is made as 0) in the scope of spatial frequency range higher limit T that is passed through and spatial frequency range lower limit B appointment or multiply by fully operation less than 1 value.Fully as mentioned above less than 1 value.

In the spatial frequency band that extracts by the application of bandpass filter (seeing through frequency band), each spatial frequency composition see through ratio (ratio of the oscillator intensity before the oscillator intensity after bandpass filter is used is used bandpass filter), shown in Figure 10 (shape that sees through the frequency band peak value has rectangle), seeing through the frequency band entire scope is constantly also can, shown in Figure 11 (seeing through the Gaussian that is shaped as of frequency band peak value), its value changes also can.In addition, the peak value shape that sees through ratio with respect to spatial frequency axle left-right symmetric also can, shown in Figure 12 (seeing through being shaped as of frequency band peak value), also can for asymmetric at the right side of the peak value distortion Gaussian different with the left side gradient.In addition, shown in Figure 13 and Figure 14 (seeing through the frequency band peak value is made of two peak values), see through the frequency band peak value and constitute by a plurality of peak values and also can.

Figure 15 is an example of the two-dimensional space frequency distribution after the bandpass filter has been used in expression to first pattern with spatial frequency distribution shown in Figure 5 figure.In Figure 15, the concentration of transverse axis, the longitudinal axis and look is represented the meaning identical with Fig. 5.As shown in figure 15, by above-mentioned operation, will remove or its oscillator intensity will be reduced by the spatial frequency composition of the particular range of spatial frequency range higher limit T and spatial frequency range lower limit B appointment corresponding to bandpass filter.

Then, the spatial frequency range higher limit T that gives of spatial frequency range lower limit B ' that Hi-pass filter is given and bandpass filter and the preferred range of spatial frequency range lower limit B describe.Under considering with the situation of random pattern of the present invention as the fundamental design use that is used to make anti-dazzle film, for the anti-dazzle film that obtains to suppress efficiently glimmering, the low spatial frequency composition of being removed or being reduced by Hi-pass filter or bandpass filter is preferably: with respect to the Pixel Dimensions on average one side of image display device (for example, under RGB three looks situation laterally arranged side by side, the Pixel Dimensions on RGB average one side separately is the mean value of long limit and minor face) the following low spatial frequency composition of about pairing spatial frequency of cycle 1/10th below, the described image display device image display device of the transparent base of non-glare treated (anti-dazzle film etc.) that has been the carrying out that obtain by the present invention of application.Thus, can suppress the flicker of image display device efficiently.

The image display device of selling on market is that example is when carrying out concrete narration, for example, be applied to the about 103 inches full HD TV of diagonal line (sharpness: level 1920 * vertical 1080 etc.) under the situation of corresponding image display device, the maximal value of the spatial frequency of the low spatial frequency composition of removing or reducing by Hi-pass filter or bandpass filter, be spatial frequency range lower limit B ' or spatial frequency range lower limit B, be preferably 0.01 μ m ^-1More than.In addition, be applied to the about 50 inches full HD TV of diagonal line (sharpness: level 1366 * vertical 768 etc.) under the situation of corresponding image display device, spatial frequency range lower limit B ' or spatial frequency range lower limit B are preferably 0.02 μ m ^-1More than.According to same investigation, be applied to the about 32 inches full HD TV of diagonal line under the situation of corresponding image display device, spatial frequency range lower limit B ' or spatial frequency range lower limit B are preferably 0.03 μ m ^-1More than.Be applied to the about 37 inches full HD TV of diagonal line under the situation of corresponding image display device, spatial frequency range lower limit B ' or spatial frequency range lower limit B are preferably 0.04 μ m ^-1More than.Be applied to the about 20 inches full HD TV of diagonal line under the situation of corresponding image display device, spatial frequency range lower limit B ' or spatial frequency range lower limit B are preferably 0.05 μ m ^-1More than.Be applied to the about 22 inches full HD TV of diagonal line under the situation of corresponding image display device, spatial frequency range lower limit B ' or spatial frequency range lower limit B are preferably 0.07 μ m ^-1More than.Like this, by suitably regulating the spatial frequency range lower limit of giving to Hi-pass filter or bandpass filter according to the sharpness and the size of the image display device of using, can generate second, third or the 4th pattern being removed or reduced with respect to the low spatial frequency composition of image display device proper range, by being fundamental design with this pattern, and, can realize suppressing the anti-dazzle film of flicker based on this fundamental design processing concaveconvex shape.

In addition, in bandpass filter, from being the relief pattern of fundamental design or the viewpoint of the flexibility (adaptability) of operation that printed patterns is processed with random pattern of the present invention, spatial frequency range higher limit T is preferably 1/ (D * 2) μ m ^-1Below.At this, D (μ m) is based on the resolution of device of random pattern of the present invention (the 3rd or the 4th pattern) processing concaveconvex shape or the resolution of the printing equipment that prints based on random pattern of the present invention.Surpass 1/ (D * 2) μ m at spatial frequency range higher limit T ^-1Situation under, machining reproducibility give concaveconvex shape well or print become the difficulty.T is more little for the spatial frequency range higher limit, and machining reproducibility is good more, so spatial frequency range higher limit T is preferably 1/ (D * 4) μ m ^-1Below, 1/ (D * 6) μ m more preferably ^-1Below.At spatial frequency range higher limit T is 1/ (D * 6) μ m ^-1Under the following situation, for example can utilize the high laser drawing apparatus of throughput rate on transparent base, to form concaveconvex shape with good machining reproducibility, therefore preferred especially.On the other hand, spatial frequency range higher limit T is big more, and easy more formation has second pattern of shorter structure of cycle, and therefore processing is reproduced and is easy to become difficulty.

The device of processing concaveconvex shape can be existing known device on transparent base, for example, can use laser drawing apparatus, accurate rotating disk etc.Using the laser drawing apparatus that the resist exposure is formed under the situation of concaveconvex shape, the spot diameter of laser is equivalent to resolution D (μ m).In addition, utilization possess front end be the accurate rotating disk of hemispheric ball end mill form the situation of concaveconvex shape and hold before use radius for the face angulation of the plane of the ball end mill of r (μ m) male and fomale(M﹠F) after according to processing and each position be the θ degree (for example, θ is 10 degree) process under the situation of concaveconvex shape in interior mode, 2 * r ÷ (sin (θ ÷ 180 π)) is equivalent to resolution D (μ m).In addition, by make metal pattern with male and fomale(M﹠F) based on random pattern of the present invention (the 3rd or the 4th pattern) and with the male and fomale(M﹠F) transfer printing of metal pattern under the situation of processing concaveconvex shape on the transparent base, the device of processing concaveconvex shape is the processing unit (plant) that uses when making the metal pattern with male and fomale(M﹠F) on transparent base.

In addition, under the situation of considering random pattern of the present invention is used as being used to make the fundamental design of anti-dazzle film, the inverse of promptly maximum cycle length 1/B of the inverse of spatial frequency range lower limit B and spatial frequency range higher limit T is that the centre of minimum cycle length 1/T is that length M ainperiod=intercycle (1/B+1/T)/2 is preferably in the scope below the above 33 μ m of 6 μ m.Mainperiod is equivalent to give to the pairing Cycle Length of spatial frequency range higher limit T (1 ÷ T) the μ m of bandpass filter and the mean value of the pairing Cycle Length of spatial frequency range lower limit B (1 ÷ B) μ m.Under the situation of Mainperiod, when processing concaveconvex shape on transparent base, be difficult for forming spatial frequency and be lower than 0.10 μ m greater than 33 μ m ^-1Small convex-concave surface shape, can not embody anti-dazzle property efficiently.In addition, under the situation of Mainperiod, when processing concaveconvex shape on transparent base, may form spatial frequency and be lower than 0.01 μ m less than 6 μ m ^-1Small convex-concave surface shape, consequently, when resulting anti-dazzle film being disposed at high meticulous image display device surperficial, may glimmer.

In addition, by maximum cycle length 1/B, minimum cycle length 1/T, and the centre of maximum cycle length 1/B and minimum cycle length 1/T be length M ainperiod intercycle (=(1/B+1/T)/2) defined following formula (1)

BandWidth(％)＝100×(1/B-1/T)/(2×MainPeriod) (1)

The BandWidth (bandwidth) of expression preferably satisfies following formula (2).

15≤BandWidth(％)≤70 (2)

Shown in above-mentioned formula (1), the numerical value that BandWidth is directly proportional for the difference with maximum cycle length 1/B and minimum cycle length 1/T.BandWidth also can be by with length M ainperiod=intercycle (1/B+1/T)/2 substitution following formula (1) and by following formula (1) ' define.

BandWidth(％)＝100×(1/B-1/T)/(1/B+1/T) (1)’

Figure 16 is the value of expression BandWidth and the figure of the peaked relation of coefficient of autocorrelation.The coefficient of autocorrelation maximal value is the maximal value of coefficient of autocorrelation.Coefficient of autocorrelation can obtain in the following way, promptly, based on Wei Na-Xin Qin (ウ イ one Na one ヒ Application チ Application) theorem, the 3rd pattern is transformed to by two-dimensional Fourier transform after the two-dimensional array in spatial frequency zone, the coefficient of each key element is carried out a square calculating, and further this is implemented inversefouriertransform.The coefficient of autocorrelation maximal value is expression and the numerical value that becomes index of the parallel mobile relevant autocorrelative intensity of self.Therefore, the coefficient of autocorrelation maximal value is high more, and in the relief pattern of processing on transparent base and printed patterns, similar pattern is easily continuous, increases with the similarity that is present near pattern, therefore easily damages the randomness of pattern.In addition, coefficient of autocorrelation maximal value shown in Figure 16 is the coefficient of autocorrelation maximal value relevant with the 3rd pattern, and be that displacement is the coefficient of autocorrelation maximal value in the above scope of 20 μ m, described the 3rd pattern is to generate behind second pattern by the bandpass filter that is shaped as rectangle of first pattern being used Mainperiod=12 μ m, being seen through the frequency band peak value error-diffusion method (about error-diffusion method, the back is described in detail) of this second pattern application error diffusion length 5 is generated.

As shown in figure 16, when BandWidth less than 15%, the coefficient of autocorrelation maximal value greatly increases as can be known; On the other hand, under BandWidth was situation more than 15%, the coefficient of autocorrelation maximal value was kept lower value.Therefore, in order to obtain and be present near the pattern that similarity is low and randomness is higher of pattern, the value of BandWidth is preferably more than 15%, more preferably surpasses 15%.

On the other hand, the result of research shows that the spatial frequency range that second pattern has is big more, has following tendency more, that is, the many compositions different by Cycle Length replenish mutually, and isolated point is easy to produce when second pattern is carried out binary conversion treatment described later.Figure 17 be expression BandWidth value, with pass through the figure of second pattern by the relation of the generation number of the isolated point of resulting the 3rd pattern of error-diffusion method binaryzation described later, and be to generate behind second pattern the 3rd pattern that the error-diffusion method (about error-diffusion method, the back is described in detail) to this second pattern application error diffusion length 5 generated by the bandpass filter that is shaped as rectangle of first pattern being used Mainperiod=12 μ m, being seen through the frequency band peak value to scheme accordingly.Among Figure 17, " the generation number of isolated point " is meant the piece (island) that exists in the 3rd pattern, be made of the continuous same color pixel (pixel) below 16.

As shown in figure 17, as can be known, surpass under the situation more than 70% at BandWidth, the generation number of isolated point presents the tendency of rapid increase; On the other hand, under BandWidth was situation below 70%, the generation number of isolated point was kept lower value.Therefore, for the machining reproducibility that makes concaveconvex shape is good, the value of BandWidth is preferably below 70%, more preferably below 65%.

As known from the above, in order to obtain the good and higher pattern of randomness of machining reproducibility, BandWidth preferably satisfies following formula (2).Satisfy the bandpass filter of following formula (2) by application, needn't utilize the reduction of the isolated point of Monte Carlo method described later to handle, just can obtain the good random pattern of machining reproducibility.

In addition, to use after the bandpass filter, similarly make the preferred constant mode of oscillator intensity, implement the processing of increase and decrease oscillator intensity according to situation in the spatial frequency distribution of the scope of spatial frequency range lower limit B～spatial frequency range higher limit T with the spatial frequency distribution of first pattern.Change by the oscillator intensity that makes spatial frequency composition smoothly, can obtain the relief pattern or the printed patterns of more level and smooth (have the concaveconvex shape of corner angle few).

The generation of (3) second patterns

Then, will based on this two-dimensional array, generate second pattern by implementing to be transformed to two-dimensional array by Inverse Discrete Fourier Transform corresponding to the resulting spatial frequency information of the operation of Hi-pass filter or bandpass filter.As the IDFT algorithm, same with above-mentioned DFT, can use generally well-known algorithm.Second pattern can have various bit depth such as 8,16,32,64.

Figure 18 is an example of second pattern that bandpass filter generated is used in expression to first a pattern shown in Figure 1 enlarged drawing.Same with Fig. 1, Figure 18 also is the view data of 12800dpi.The spatial frequency range lower limit B and the spatial frequency range higher limit T that give to bandpass filter are respectively 0.043 μ m ^-1, 0.059 μ m ^-1

In addition, when generating second pattern, the maximal value of the two-dimensional array that is obtained by IDFT and minimum value are according to the mode by the maximal value of the bit depth defined of second pattern that will generate, minimum value of the corresponding respectively to substitution that converts.Promptly, when the maximal value of the two-dimensional array key element that calculates by IDFT be made as Imax, when minimum value is made as Imin, under the situation of the pattern that key element value Ix is transformed to 8 (0-255), the value of each pixel institute substitution of pattern is calculated by 255 * (Ix-Imin) ÷ (Imax-Imin).The view data of above-mentioned Figure 18 is resulting by carrying out this conversion exactly.

More than, the example that the application of Hi-pass filter by utilizing DFT or bandpass filter is generated second method of patterning is narrated, but also can generate second pattern by the method beyond it.For example, use is disposed the plate of opening and this plate is carried out Fourier transform by optical means as first pattern, also can access second pattern.Particularly, prepare the spatial frequency filtering optical system that 2 pieces of lens by the focus unanimity constitute, with the focus face of first pattern arrangement in first piece of lens.At this moment, on the face (Fourier plane) of the focus unanimity of two pieces lens, can obtain the spatial frequency distribution of image.In this Fourier plane, carry out spatiality by the transmitance that makes light and change, the spatial frequency of desired scope is seen through.

The output image of institute's filtering can obtain on the focus face of the opposition side of the Fourier plane of second piece of lens.For example, when the mode that sees through in Fourier plane according to the central part of opening only disposed plate, only the low spatial frequency composition of above-mentioned image was obtained as output image.On the contrary, when with the central part shading of opening, only the high spatial frequency composition is obtained as output image.Therefore, by on Fourier plane with core and its peripheral part shading, can on the focus face of the opposition side of second piece of lens, obtain having second pattern as the purpose spatial frequency distribution.

(generation of the 3rd pattern)

In the present invention, by second pattern that obtains is as mentioned above used the tuning method of shivering, generate the 3rd pattern of the information that is transformed to discretize.The 3rd pattern is homogeneity and randomness excellent pattern, gives the fundamental design of relief pattern or fundamental design that conduct be used to give printed patterns to various display devices with component parts as being used for, and can preferably adopt." information of discretize " is also referred to as numerical data usually, and the information of handling on computers in most of the cases is the information of discretize.As the information of discretize, can enumerate the view data that data bitmap etc. can be handled on computers; And floating numbers with various bit depth such as 128,64,32,16; Symbol is perhaps arranged or do not have integer of symbol etc.

In addition, the meaning of " to the conversion of the information of discretize " be meant with continuous function be transformed to discrete expression, with the information that simulated data is transformed to numerical data, the information conversion of the discretize that maybe will express with more progression is the discretize expressed with progression still less, and comprise the conversion of digital signal conversion for the digital signal expressed with bit depth still less.As example, enumerate the example of for example cosine function of continuous function being expressed discretely and will be example of progression 8 integers still less etc. by the information conversion of more 32 floating point expressions of progression to the conversion of the information of discretize.

The 3rd pattern is preferably pattern information, that be binaryzation that is transformed to by the secondary discretize.This situation etc. that comprises the resist workpiece that utilizes laser drawing apparatus etc. in the concavo-convex processing that with random pattern of the present invention is fundamental design is utilized the two-value of whether shining based on laser down, generates the resist pattern.

The 3rd method of patterning as the 3rd pattern, the particularly binaryzation of the information that obtains being transformed to discretize from second pattern in the present invention, adopts the tuning method of shivering.The tuning method of shivering is simulated data to the conversion of numerical data or is used for the position speed of numerical data and one of the method for the conversion of bit depth, can be used as the method location of digital signal processing.Known method that irrelevance by giving the error of random signals such as rectangular probability density function and triangle probability density function when reducing signal discrete arranged or the pattern the whole bag of tricks such as tuning method, error-diffusion method of shivering.

In above-mentioned, also can obtain the more excellent random pattern of homogeneity in the present invention, thus, can form and moire fringe be difficult for to take place or show uneven and repeat the relief pattern or the printed patterns of decorative pattern, in addition, can expect to suppress the effect of the fluctuation of local average luminance brightness, in addition, might can suppress the generation of the fine pattern of processing difficulties by the optimization of matrix, therefore, preferably adopt error-diffusion method as shivering tuning method.Being characterized as of error-diffusion method, the error diffusion that produces when making discretize is to periphery.

About the summary of the algorithm of error-diffusion method, the situation that is transformed to the black white bitmap of 12 tone with the gray bitmap with 8 256 tones is that example describes.At present, the brightness values that has of transforming object pixel (pixel) is made as 64.Under the situation of the black white bitmap that this pixel is transformed to 12 tone, need be transformed under 8 with the expressed white of brightness values 255 or with the expressed black of brightness values 0.Usually be transformed to more approaching value.Therefore, brightness values is 64 pixel owing to than 255 more near 0, therefore be transformed to the value (that is black) corresponding to 0.At this moment, by conversion, when comparing, produce-64 brightness values error in the image after conversion with the image of 8 tones.This meaning is that the summation of the luminance brightness of image has reduced 64.In error-diffusion method, offsetting the mode of-64 the brightness values error produced, according to the flexible strategy of decision in advance, with around the brightness values of pixel change.By this operation all pixels are all repeated, carry out binaryzation.

About the method for weighting, in image processing field, known have several preferred matrixes.For example, Floyd ﹠amp; Steinberg; Jarvis, Judis and Nink; Stucki; Burks; Stevenson ﹠amp; Arche; Sierra 3Line; Sierra 2L Line; Conducts such as Sierra Filter Lite have the matrix of preferred weighting by known.

Figure 19 is used for the figure that the weighting to the diffusion of the mapping fault of above-mentioned illustrative matrix describes.As an example of matrix, with Floyd ﹠amp; Steinberg is that example describes, and pixel A is the transforming object pixel.As mentioned above, conversion by pixel A (brightness values from 64 to 0 conversion), produced in the image after conversion under the situation of-64 brightness values error, according to the mode of offsetting this brightness values error, with the brightness values of 4 pixels of adjacency with 7: 1: 5: 3 weighting is changed.That is, the brightness values with 4 pixels of adjacency increases (7/16) * 64, (1/16) * 64, (5/16) * 64, (3/16) * 64 respectively.In addition, the pixel B of band diagonal line is represented the pixel after binary conversion treatment is finished.In addition, recording and narrating pixel for " 0 " is that not make the flexible strategy of error diffusion be zero pixel.

Second pattern that application by bandpass filter is obtained is used the example of resulting the 3rd pattern of the corresponding error-diffusion method of matrix shown in Figure 19, is shown in Figure 20～27.The 3rd pattern shown in Figure 20～27 all is according to generating as 8 resulting second patterns shown in Figure 28 of gray level image, and is made of 1 black white image data.More specifically, the 3rd pattern shown in Figure 20～27 is by utilizing the error-diffusion method of various matrixes to carry out the pattern after the binaryzation in second pattern shown in Figure 28, and described second pattern shown in Figure 28 is by being following formula (I) and (II) to the first pattern application space frequency range lower limit B and spatial frequency range higher limit T

B＝1/(MainPeriod*(1+BandWidth/100))

(I)

T＝1/(MainPeriod*(1-BandWidth/100))

And the bandpass filter that is shaped as rectangle that sees through the frequency band peak value obtains, and described first pattern is by utilizing simulation random number series that randomizer subtraction algorithm by Knuth generates and that have a value of 0～1 to make the pattern of 8 the bit image in 1.024mm four directions under the sharpness of 12800dpi.And, Mainperiod=12 (μ m), BandWidth=20 (%).In addition, with regard to Figure 20～27, it is its local amplification to be represented from the 3rd pattern that is generated for the ease of the feature of grasping image.

At this,, except that the tuning method of shivering of the present invention, it is also known for threshold method as the method for carrying out binaryzation.In the threshold method, in gray scale index (brightness values), set specific threshold value, give white (or black), give black (or white), carry out binaryzation thus the pixel below the threshold value to the pixel (pixel) that surpasses threshold value.As follows, compare with threshold method, the tuning method of shivering is realized, the especially conversion to the information of discretize that realizes of error-diffusion method can further reduce the low spatial frequency composition, therefore, obtain on the higher pattern this point of homogeneity favourable.

Figure 29 is the spatial frequency distribution of the 3rd pattern that passes through the corresponding error-diffusion method of various matrixes institute binaryzation shown in Figure 20～27, the figure that compares with the spatial frequency distribution of the pattern of passing threshold method institute binaryzation.As shown in figure 29, carry out in the passing threshold method under the situation of binaryzation, resulting pattern illustrates than higher oscillator intensity in the low spatial frequency zone.On the other hand, under the situation of application error diffusion method, when adopting arbitrary matrix, the low spatial frequency composition is further reduced.Therefore, by the application of error-diffusion method, can access the higher pattern of homogeneity.In addition, under the situation that random pattern of the present invention is used as being used to make the fundamental design of anti-dazzle film, being reduced on the anti-dazzle film that obtains suppressing efficiently to glimmer of low spatial frequency composition is favourable.In addition, the pattern by threshold method institute binaryzation of Figure 29 is: to second pattern shown in Figure 28 by will be under the state that is made as threshold value with intermediate value 127 than it figure of making of big value is made as white, value below with it is made as black binaryzation.

Like this, by the application of the corresponding error-diffusion method of generally well-known error diffusion matrix shown in Figure 19, can obtain the 3rd higher pattern of homogeneity.But, in the method that will make, have the tendency that takes place as the non-existent isolated pixel of the group more than the stated number (below, be called " isolated point ") with color pixel according to the 3rd pattern of these error diffusion matrix institute binaryzations morely.At this, " isolated point " is meant piece (island) existing in the pattern of binaryzation, that be made of the continuous same color pixel (pixel) below 16.Have at the 3rd pattern under the situation of many isolated points, can exist is the following pieces (island) of 4 pixels on one side, for example, CTP method and comprise the operation of Wet-type etching or rotating disk processing etc. utilizes the precision that can have relatively high expectations in the concavo-convex processing of this pattern and the printing processing, can hinder machining reproducibility sometimes.

The figure that Figure 30 is the generation number of the isolated point that takes place when the application by the corresponding error-diffusion method of generally well-known error diffusion matrix is generated the 3rd pattern, compare with the situation of utilizing threshold method to generate.The ratio of the generation number of the isolated point that illustrated numeric representation takes place when generating the pattern of binaryzation with respect to the passing threshold method.As shown in figure 30, at the Stevenson ﹠amp of the generation frequency minimum of isolated point; In the matrix of Arche, the generation number also is 27 times of threshold method, is using Floyd﹠amp; Under the situation of the matrix of Steinberg, also reach 155 times.

What present inventors studied with keen determination found that, in order to suppress the generation number of isolated point, as error diffusion matrix, preferred use does not comprise the matrix of short-range error diffusion.

Figure 31～39 are that expression diffusion length is 1,2,3,4,5,6 respectively, the figure of the example of the error diffusion matrix of 3+4,4+5 and 3+4+5.Same with Figure 19, these figure are the diffusion-weighted figure of expression mapping fault.Diffusion length (error diffusion distance) to be meant in order offsetting by transforming object pixel (pixel A) and to be used to change the pixel of brightness values and the distance of transforming object pixel to the brightness values error that conversion produced of white or black that the meaning of " diffusion length 1 " is the pixel and the transforming object pixel adjacency (with reference to Figure 31) of change brightness values.The meaning of " diffusion length 2 " is that second pixel that will begin to count from the transforming object pixel is made as the pixel (accompanying a pixel between pixel that changes brightness values and transforming object pixel) (with reference to Figure 32) that changes brightness values.Following diffusion length too.In addition, Figure 37 " diffusion length is the matrix of 3+4 " is the synthetic of " diffusion length is 3 matrix " shown in Figure 33 and " diffusion length is 4 matrix " shown in Figure 34.Figure 38 and Figure 39 are too.

In addition, will be shown in Figure 40～48 respectively by the example of using the 3rd pattern that the corresponding error-diffusion method of matrix shown in Figure 31～39 obtains.Employed second pattern is a pattern shown in Figure 28.In addition, amplify the part and the figure of expression from the 3rd pattern that is generated for the ease of the feature of grasping image Figure 40～48th.In addition, Figure 49 is the figure that compares with the situation that the passing threshold method generates by the generation number of using the isolated point that takes place when the corresponding error-diffusion method of error diffusion matrix shown in Figure 31～39 generates the 3rd pattern.The ratio of the generation number of the isolated point that illustrated numeric representation takes place when generating the pattern of binaryzation with respect to the passing threshold method.

As shown in figure 49, as can be known, be under 1 the situation, to compare in error diffusion distance with threshold method, the isolated point of the number up to 247 times takes place, along with specification error diffusion length significantly, number takes place reduce.As can be known, particularly surpass under 1 the situation in the error diffusion distance, the quantity of isolated point sharply reduces.By result shown in Figure 49 as can be known, in order to suppress the generation of isolated point more efficiently, the error diffusion distance is preferred to surpass 1 (that is, make the mapping fault diffusion in the scope that surpasses 1 pixel, below same), more preferably more than 2, more preferably more than 3.In addition, the upper limit of error diffusion distance is not made specific limited, for example, is below 6.Especially, utilize the pattern that matrix generated with the error diffusion distance more than 3, the range of work is extensive and can expect good flexibility (adaptability) of operation.

Figure 50 is the figure that the spatial frequency distribution that will implement the pattern of the spatial frequency distribution of the 3rd pattern of Figure 40～48 of binaryzation and passing threshold method binaryzation by the corresponding error-diffusion method of error diffusion matrix shown in Figure 31～39 compares.Pattern by this threshold method institute binaryzation is identical with the pattern of Figure 29.As shown in Figure 50, under the situation of using arbitrary error diffusion matrix, compare, the amplitude of low spatial frequency composition is reduced with threshold method.

(generation of the 4th pattern)

In the present invention,, further implement to make the operation of isolated point minimizing, generate the 4th pattern and also can being transformed to (binaryzation) above-mentioned the 3rd pattern by the information of secondary discretize.By the 3rd pattern being transformed to the 4th pattern that isolated point has reduced, can further improve with the 4th pattern is the relief pattern of fundamental design or the machining reproducibility of printed patterns processing.The pattern of the employed binaryzation of generation of the 4th pattern also can be the pattern of passing threshold method institute binaryzation, but when considering the homogeneity of pattern, preferred the 3rd pattern that uses by tuning method institute binaryzations of shivering such as error-diffusion methods.But, as mentioned above, generate under the situation of second pattern in the bandpass filter that satisfies above-mentioned formula (2) by application, may not need the minimizing of this isolated point to handle.

As the operation that above-mentioned isolated point is reduced, it is the method that black or white pixel move to homochromy piece (island) that preferred use makes the existing isolated point of the 3rd pattern by Monte Carlo method.Monte Carlo method is based on the general name of the method that random number simulates.As the disposal route of isolated point, the method for eliminating isolated point simply is the simplest.But, in Flame Image Process, when adopting this straightforward procedure, there is the average light brightness value to change sometimes in the part, this can involve the increase of low spatial frequency composition.Monte Carlo method can bring influence for partly average light brightness yet, is the effective method of handling isolated point.Below, with reference to Figure 51 the concrete example of the disposal route of the isolated point of Monte Carlo method realization is described.

At first, judge whether object pixel (pixel) is " isolated point "." isolated point " at the concrete example of this explanation is defined as, in the most adjacent 8 pixels on every side, the number that is positioned at (homochromy) pixel of the grade identical with object pixel is below 2.For example, be under the situation of black in object pixel, if the number of black picture element is below two in the most adjacent 8 pixels, then be judged to be isolated point.White pixel too.Then, the pixel that the is judged to be isolated point pixel of being selected by random number in empty neighbor is moved.

For example, in Figure 51 (a), be under the situation of black in object pixel, in the most adjacent 8 pixels only a pixel be black, therefore be judged to be isolated point, object pixel is moved in the pixel of being selected by random number in the most adjacent empty 7 pixels.In addition, in Figure 51 (b), be under the situation of black in object pixel, therefore two pixels are black in the most adjacent 8 pixels, are judged to be isolated point, and object pixel is moved in the pixel of being selected by random number in the most adjacent empty 6 pixels.In Figure 51 (c), be under the situation of black in object pixel, three pixels are black in the most adjacent 8 pixels, therefore are not judged to be isolated point, do not move it.

By repeating the operation of aforesaid Monte Carlo method, can reduce isolated point efficiently.If the operation of Monte Carlo method repeats for example 10～60 degree, then see through the value of spatial frequency of the spatial frequency composition of bandpass filter, by being scaled Cycle Length and between 3 pixels～6 pixels the time, almost can't check isolated point, can obtain expecting the adaptive pattern of well processed.

Figure 52 (a)～(f) is the figure of expression based on the variation of the 4th pattern of Monte Carlo method number of applications.Pattern shown in Figure 52 (a)～(f) is that the 3rd pattern shown in Figure 44 (diffusion length 5) is used respectively that 0,4,8,20,40 and 60 time Monte Carlo method is handled isolated point and the pattern that obtains.In addition, Figure 53 is the figure of relation of the generation number of expression Monte Carlo method number of applications and isolated point.Same with Figure 30 and Figure 49, the isolated point generation number of Figure 53 is than the ratio that is the generation number of the isolated point that takes place during with respect to the pattern that generated binaryzation by the second pattern passing threshold method shown in Figure 28.Like this,, isolated point can be reduced, the 4th pattern of the more excellent flexibility (adaptability) of operation of expectation can be generated by the repeated application Monte Carlo method.

The generation example of the 4th above-mentioned pattern, as second pattern, use is used the pattern that bandpass filter generated to first pattern, but using under the situation of using second pattern that Hi-pass filter generated, also with the situation of bandpass filter similarly, minimizing by binaryzation and isolated point is handled, and can reduce the low spatial frequency composition, can obtain the 4th pattern of flexibility (adaptability) of operation excellence.

More than shown in, second pattern is used the tuning method (especially error-diffusion method) of shivering generate the 3rd pattern and the 3rd pattern used Monte Carlo method and generate the 4th method of patterning, when generating second pattern, under the situation of not using the bandpass filter that satisfies above-mentioned formula (2), therefore the pattern that also can obtain excellent in uniformity (the low spatial frequency composition is minimized), randomness excellence and isolated point has been reduced is one of preferred embodiment.

The relief pattern of＜use random pattern or the processing of printed patterns 〉

The above-mentioned random pattern of the present invention that obtains (the 3rd pattern or the 4th pattern) can preferably be given the fundamental design of relief pattern with component parts or use as the fundamental design of giving printed patterns for example display devices such as anti-dazzle film, diffuser plate, light diffusing sheet, light guide plate as being used for, thus, can form the relief pattern or the printed patterns of homogeneity and randomness excellence.Random pattern of the present invention is owing to be the pattern that is transformed to the information of discretize, and therefore the device that uses when above-mentioned relief pattern of processing or printed patterns need be transformed under the situation of fundamental design of information of discretize, and is effective especially.

For example, with regard to relief pattern to display device with regard to the giving of component parts, can by with random pattern of the present invention be fundamental design and on transparent base the processing concaveconvex shape carry out.Particularly, for example, can be by the method for the following stated, based on random pattern processing concaveconvex shape of the present invention.Employed device can be existing known device when processing concaveconvex shape on transparent base, can use for example laser drawing apparatus, laser processing device, automatic carving device, accurate rotating disk etc.As laser processing device, can use the various processing unit (plant)s of selling as for example laser-marking device, laser engraving machine, laser machine etc.

For example, using laser drawing apparatus etc. to have under the situation of processing unit (plant) of resist operation, the information of the discretize that random pattern has is preferably the information by the secondary discretize.In this device, further use under the situation of the two-dimensional array processing concaveconvex shape of pressing the secondary discretize, as long as like that following.At first, based on luminance brightness information etc., random pattern is transformed to two-dimensional array g[x, y].At this, x, y represent the position coordinates shown in each key element of two-dimensional array.Secondly, confirm pressing the two-dimensional array g[x of secondary discretize, y] all key elements value of storing.At this, suppose by the two-dimensional array that operates in to store 0 or 1 by the secondary discretize.In the processing of concaveconvex shape, for example, at the key element g[a1 of the two-dimensional array corresponding, b1 with ad-hoc location x=a1, y=b1] value of being stored is under 1 the situation, in processing unit (plant), to the coordinate irradiating laser corresponding to a1, b1, to form recess.Be under 0 the situation in the value of being stored, not to the coordinate irradiating laser of correspondence.By all key elements are repeated this operation, can obtain concaveconvex shape by random pattern.Have only under the situation of the intensity that forms recess on the processing object, at laser by the irradiation formation recess of laser.When the weak strength of laser, describe to make resist sensitization by laser, after resist is developed, form recess by etching, process concaveconvex shape thus and also can.

In addition, use as processing unit (plant) under the situation of numerical control cutting processing unit (plant),, can use the pattern that is transformed to the information of progression (or number of significant digit) the institute discretize of its processing unit (plant) appointment as random pattern.When with random pattern of the present invention being the formation of printed patterns of fundamental design, can use existing known printing equipment.

In addition, under the situation that random pattern of the present invention is made of the two-dimensional array of the information of discretize, in the processing of carrying out based on the value that is stored in this two-dimensional array, can be used for processing according to these values of characteristic conversion of processing unit (plant).For example, under the situation of laser machine and laser engraving machine, also can replace with the laser radiation number of times.Under the situation of the processing unit (plant) of the degree of depth of the sort of control cutter of accurate rotating disk, the also variable amount that is changed to corresponding to the cutter amount of being pressed into.So that be example, the conversion of value is specifically described in order to the situation of the two-dimensional array of 8 tone institute discretizes.At this moment, two-dimensional array g[x, y] be assumed to and take 0～255 value.The intensity of anti-dazzle property can be controlled by the difference of height of concaveconvex shape.To the formula that carries out conversion corresponding to the amount of the cutter amount of being pressed into by the difference of height of necessity with at two-dimensional array g[x, y] maximal value and the minimum value decision of the value of being stored.Difference of height to be made as under the situation of 1 μ m, when the cutter amount of being pressed into from the surface of smooth processing object among coordinate x, the y is made as z, by utilizing

Z=(g[x, y]-minimum value)/(maximal value-minimum value) * difference of height

Calculate, decide the cutter amount of being pressed into.In the object lesson described herein, by being made as

z＝(g[x，y]-0)/(255-0)×1μm

Calculate, can obtain the cutter amount of being pressed into is z.That is,, y at g[x] value be under 255 the situation, the cutter amount of being pressed into z to be made as 1 μ m, at g[x, y] value be under 0 the situation, the cutter amount of being pressed into z to be made as 0 μ m.By to being stored in two-dimensional array g[x, y] all key elements carry out this calculating, form concaveconvex shape.Under situation, under the prerequisite of the relation of in advance having confirmed irradiation number of times and working depth, determine to shine number of times according to the mode corresponding to the value of above-mentioned z of becoming and get final product by laser radiation number of times controlling depth.

As mentioned above, be that the information of the carving depth of transparent base is reflected on the concaveconvex shape or the information decision that utilizes random pattern to have forms recess and still do not form recess by the information conversion that random pattern is had, carry out concaveconvex shape processing.In addition, be subjected to the restriction of the resolution of processing unit (plant), under the excessive situation of difference of height,, difference of height reduced also can by whole of etching.In addition, as the method that on transparent base, forms concaveconvex shape, also can be directly on transparent base, to implement above-mentioned method for processing, also can preferably adopt following method, promptly, on metal pattern, form after the concaveconvex shape based on pattern with said method, by with the concaveconvex shape transfer printing of metal pattern on transparent base, on transparent base, form concaveconvex shape based on pattern.

[example]

Below, the present invention will be described in more detail for embodiment, but the present invention is not limited to these embodiment.

＜embodiment 1 〉

Random pattern shown in Figure 54 (the 4th pattern) is generated.The 4th pattern shown in Figure 54 is the pattern by the 32.768mm four directions that sharpness generated of 12800dpi, and Figure 54 is the figure that has wherein cut out the 1.024mm four directions.The 4th pattern is to generate by the Monte Carlo method to the further repeated application of the 3rd pattern 60 times, described the 3rd pattern is to be that 4 the corresponding error-diffusion method of error diffusion matrix shown in Figure 34 generates the second pattern binaryzation by application error diffusion length, described second pattern is to use above-mentioned formula (I) to reach (II) (MainPeriod=12 (μ m) respectively by first pattern is used, BandWidth=20%) representation space frequency range lower limit B and spatial frequency range higher limit T and the bandpass filter that is shaped as Gaussian that sees through the frequency band peak value obtain, and described first pattern is by utilizing simulation random number series that randomizer subtraction algorithm by Knuth generates and that have a value of 0～1 to make the pattern of 8 the bit image in 32.768mm four directions under the sharpness of 12800dpi.

＜comparative example 1 〉

Except that utilizing threshold method (threshold setting is a gray scale index (brightness values) 127) to carry out the binaryzation, the 3rd pattern that generates with the foregoing description 1 similarly generates the pattern shown in Figure 55.

Figure 56 is the figure that the spatial frequency distribution with the pattern of the spatial frequency distribution of the random pattern of the embodiment shown in Figure 54 1 and the comparative example shown in Figure 55 1 compares.By Figure 56 as can be known, in the pattern of Figure 54 of application error diffusion method, the low spatial frequency composition further reduces, and homogeneity is more excellent.In addition, calculated the coefficient of autocorrelation maximal value of the pattern of the random pattern of embodiment 1 and comparative example 1, from its result as can be known, when the coefficient of autocorrelation maximal value with the pattern of comparative example 1 is made as 1, the coefficient of autocorrelation maximal value of the random pattern of embodiment 1 is about 0.95, has fully high randomness.

＜comparative example 2 〉

Do not use bandpass filter and error-diffusion method, only generate the pattern shown in Figure 57 by disposing many points randomly.Pattern shown in Figure 57 is by every 1mm ²Be arranged with the pattern of 5000 points that are equivalent to the about 15 μ m of diameter.In order to be made as the pattern a bit that as far as possible distributes equably, by setting triangular crystal lattice corresponding to the dot density of having set, from its lattice-site make a little centre coordinate X and Y respectively with respect to the lattice displacement of the triangular crystal lattice that sets, generate pattern.In addition, the program code of the C# shown in the decision of the coordinate after displacement use is following (being that the language specification is by regulations such as " JIS X 3015 program language C# " by the program language of Microsoft company exploitation).In this function, give 0.3 * 15 μ m by coordinate figure (X or Y) and Deviation to the lattice-site that is made as the displacement of Average institute, make position displacement randomly.At this moment, simulation random number (" the RandomFunction () j " of C# program code) is to give numerical value 607 as kind and obtain by SIMDoriented Fast Mersenne Twister program, the SFMT ver1.3.3 that the research group by Hiroshima University is proposed.

(program code of the C# that comparative example 2 uses)

//cx, cy represent x coordinate, the y coordinate of the dot center newly described.

//px, py represent x coordinate, the y coordinate of the triangular crystal lattice point that sets.

//pD：0.3

//Core Size: the diameter of point

cX＝Normal?Random(px，pD*Core?Size)

cY＝Normal?Random(py，pD*Core?Size)

//: the regulator of random number

//RandomFunction (): the function that recovers random number

//RandomFunctionValueMax (): the peaked function of the value that the recovery random number is got

//Math:NET Framework Math class method storehouse

public?double?NormalRandom(double?Average，double?Deviation)

{

double?buff＝0；

buff＝Deviation*Math.Sqrt(-2*Math.Log(((double)RandomFunction

()/(double)RandomFunctionValueMax())))*Math.Sin(2*Math.PI*((doub

le)RandomFunction()/(double)RandomFunctionValueMax()))+Average；

if(buff＜0){buff＝0；}；

return?buff；

}

Figure 58 is the figure that the spatial frequency distribution with the pattern of the spatial frequency distribution of the random pattern of the embodiment shown in Figure 54 1 and the comparative example shown in Figure 57 2 compares.By Figure 58 as can be known, the low spatial frequency composition occurred in the pattern of Figure 57, on the other hand, in the pattern of the Figure 54 that has used bandpass filter and error-diffusion method, the low spatial frequency composition further reduces morely, and homogeneity is more excellent.

＜embodiment 2 〉

Generated the random pattern shown in Figure 59 (the 4th pattern).Random pattern shown in Figure 59 is the pattern with the 32.768mm four directions that sharpness was generated of 12800dpi, and Figure 59 is the figure that has wherein cut out the 1.024mm four directions.The 4th pattern shown in Figure 59 is to generate by the Monte Carlo method to the further repeated application of the 3rd pattern 60 times, and described the 3rd pattern generates behind second pattern application error diffusion method and carries out binaryzation and generate by first pattern being used bandpass filter.Employed first pattern is 8 the bit image in 32.768mm four directions under the sharpness of 12800dpi, by to having the two-dimensional array PIXCEL[x of 8 bit depth, y] substitution PIXCEL[x, y]=R[x+y * Imagewidth] * 255 and generate.At this, x, y are the pixel coordinates in the image, and Imagewidth is the pixel wide of x coordinate.As array R[], the simulation random number series that uses the randomizer subtraction algorithm of the Knuth that takes the value between 0.0 and 1.0 that is generated by " NETFramework2.0 class method storehouse " contained Random class NextDouble method to draw.As bandpass filter, usage space frequency range lower limit B is 0.045 μ m ^-1, spatial frequency range higher limit T is 0.080 μ m ^-1, and see through the bandpass filter that the frequency band peak value has the more steep asymmetrical shape of the inclination of low spatial frequency side.In addition, as error diffusion matrix, using diffusion length shown in Figure 33 is that 3 error diffusion matrix and diffusion length shown in Figure 34 are 4 error diffusion matrix by 0.4: 0.6 the error diffusion matrix that ratio synthesized (Figure 33 * 0.4+ Figure 34 * 0.6).The spatial frequency distribution of the random pattern shown in Figure 59 is shown in Figure 60.As can be known, the low spatial frequency composition is reduced, excellent in uniformity.

＜embodiment 3 〉

Generated the random pattern shown in Figure 61 (the 4th pattern).Random pattern shown in Figure 61 is the pattern in the 32.768mm four directions that generated under the sharpness of 12800dpi, and Figure 61 is the figure that has wherein cut out the 1.024mm four directions.Random pattern shown in Figure 61 generates the Monte Carlo method of the further repeated application of the 3rd pattern 60 times, and described the 3rd pattern generates behind second pattern application error diffusion method and carries out binaryzation and generate by first pattern being used bandpass filter.Employed first pattern is 8 the bit image in 32.768mm four directions under the sharpness of 12800dpi, by to having the two-dimensional array PIXCEL[x of 8 bit depth, y] substitution PIXCEL[x, y]=R[x+y * Imagewidth] * 255 and generate.At this, x, y are the pixel coordinates of image, and Imagewidth is the pixel wide of x coordinate.As array R[], the simulation random number series that uses the randomizer subtraction algorithm of the Knuth that takes the value between 0.0 and 1.0 that is generated by " NETFramework2.0 class method storehouse " contained Random class NextDouble method to draw.As bandpass filter, usage space frequency range lower limit B is 0.055 μ m ^-1, spatial frequency range higher limit T is 0.100 μ m ^-1, and see through the bandpass filter that is shaped as the Gaussian function type of frequency band peak value.In addition, as error diffusion matrix, using diffusion length shown in Figure 34 is that 4 error diffusion matrix and diffusion length shown in Figure 35 are 5 error diffusion matrix by 0.9: 0.1 the error diffusion matrix that ratio synthesized (Figure 34 * 0.9+ Figure 35 * 0.1).The spatial frequency distribution of the random pattern shown in Figure 61 is shown in Figure 62.As can be known, the low spatial frequency composition is reduced, excellent in uniformity.

＜embodiment 4 〉

Generated the random pattern shown in Figure 63 (the 4th pattern).Random pattern shown in Figure 63 is the pattern in the 32.768mm four directions that generated under the sharpness of 12800dpi, and Figure 63 is the figure that has wherein cut out the 1.024mm four directions.Random pattern shown in Figure 63 generates the Monte Carlo method of the further repeated application of the 3rd pattern 60 times, and described the 3rd pattern generates behind second pattern application error diffusion method and carries out binaryzation and generate by first pattern being used Hi-pass filter.Employed first pattern by make equalization point directly be 8 μ m o'clock with 10000/mm ²Density distribute randomly and generate.At this moment, in order to be made as the pattern a bit that as far as possible distributes equably, by setting triangular crystal lattice corresponding to the dot density of having set, and from its lattice-site make a little centre coordinate X and Y respectively with respect to the lattice displacement of the triangular crystal lattice that sets, generate pattern.In addition, the identical program code that the decision use of the coordinate after the displacement and comparative example 2 uses, and similarly carry out.

As Hi-pass filter, usage space frequency range lower limit B ' is 0.067 μ m ^-1Hi-pass filter.In addition, as error diffusion matrix, using diffusion length shown in Figure 34 is that 4 error diffusion matrix and diffusion length shown in Figure 35 are 5 error diffusion matrix by 0.9: 0.1 the error diffusion matrix that ratio synthesized (Figure 34 * 0.9+ Figure 35 * 0.1).

＜comparative example 3 〉

Except that utilizing the threshold method binaryzation, similarly to Example 4, be created on the local pattern of expression among Figure 64.

Figure 65 is the figure that the spatial frequency distribution with the spatial frequency distribution of the pattern shown in Figure 63 and the pattern shown in Figure 64 compares.By Figure 65 as can be known, the pattern of Figure 63 of application error diffusion method, the low spatial frequency composition further reduces, and homogeneity is more excellent.

The generation and the evaluation of the pattern that the application of＜Hi-pass filter is carried out 〉

By method shown below, generate pattern 1～15.

(1) pattern 1: to by make equalization point directly be 24 μ m o'clock with 1111/mm ²Density distribute randomly and local pattern A shown in Figure 66 of generating, application space frequency range lower limit B ' is 0.07 μ m ^-1Hi-pass filter generate second pattern after, carry out binaryzation by establishing 127 threshold methods for threshold value, obtain pattern 1.Figure 67 is the local figure that amplifies expression pattern 1.In addition, when the generation of above-mentioned first pattern, adopt the identical method of using with embodiment 4 of first pattern, realize the homogenising that point distributes.

(2) pattern 2: at employed second pattern of the generation of pattern 1, using diffusion length shown in Figure 34 is that 4 error diffusion matrix and diffusion length shown in Figure 35 are 5 the error diffusion matrix error-diffusion method by 0.9: 0.1 the error diffusion matrix that ratio synthesized (Figure 34 * 0.9+ Figure 35 * 0.1), obtains the pattern 2 as the 3rd pattern.Figure 68 is the local figure that amplifies expression pattern 2.

(3) pattern 3: the Monte Carlo method to pattern 2 repeated application 60 times obtains the pattern 3 as the 4th pattern.Figure 69 is the local figure that amplifies expression pattern 3.

(4) pattern 4: use by make equalization point directly be 20 μ m o'clock with 1600/mm ²Density distribute randomly and the local first pattern B shown in Figure 70 of generating, in addition, similarly obtain pattern 4 with pattern 1.Figure 71 is the local figure that amplifies expression pattern 4.

(5) pattern 5: at employed second pattern of the generation of pattern 4, using diffusion length shown in Figure 34 is that 4 error diffusion matrix and diffusion length shown in Figure 35 are 5 the error diffusion matrix error-diffusion method by 0.9: 0.1 the error diffusion matrix that ratio synthesized (Figure 34 * 0.9+ Figure 35 * 0.1), obtains the pattern 5 as the 3rd pattern.Figure 72 is the local figure that amplifies expression pattern 5.

(6) pattern 6: the Monte Carlo method to pattern 5 repeated application 60 times obtains the pattern 6 as the 4th pattern.Figure 73 is the local figure that amplifies expression pattern 6.

(7) pattern 7: use by make equalization point directly be 16 μ m o'clock with 2500/mm ²Density distribute randomly and the local first pattern C shown in Figure 74 of generating, in addition, similarly obtain pattern 7 with pattern 1.Figure 75 is the local figure that amplifies expression pattern 7.

(8) pattern 8: at employed second pattern of the generation of pattern 7, using diffusion length shown in Figure 34 is that 4 error diffusion matrix and diffusion length shown in Figure 35 are 5 the error diffusion matrix error-diffusion method by 0.9: 0.1 the error diffusion matrix that ratio synthesized (Figure 34 * 0.9+ Figure 35 * 0.1), obtains the pattern 8 as the 3rd pattern.Figure 76 is the local figure that amplifies expression pattern 8.

(9) pattern 9: the Monte Carlo method to pattern 8 repeated application 60 times obtains the pattern 9 as the 4th pattern.Figure 77 is the local figure that amplifies expression pattern 9.

(10) pattern 10: use by make equalization point directly be 12 μ m o'clock with 4444/mm ²Density distribute randomly and the local first pattern D shown in Figure 78 of generating, in addition, similarly obtain pattern 10 with pattern 1.Figure 79 is the local figure that amplifies expression pattern 10.

(11) pattern 11: at employed second pattern of the generation of pattern 10, using diffusion length shown in Figure 34 is that 4 error diffusion matrix and diffusion length shown in Figure 35 are 5 the error diffusion matrix error-diffusion method by 0.9: 0.1 the error diffusion matrix that ratio synthesized (Figure 34 * 0.9+ Figure 35 * 0.1), obtains the pattern 11 as the 3rd pattern.Figure 80 is the local figure that amplifies expression pattern 11.

(12) pattern 12: the Monte Carlo method to pattern 11 repeated application 60 times obtains the pattern 12 as the 4th pattern.Figure 81 is the local figure that amplifies expression pattern 12.

(13) pattern 13: use by make equalization point directly be 8 μ m o'clock with 10000/mm ²Density distribute randomly and the local first pattern E shown in Figure 84 of generating, in addition, similarly obtain pattern 13 with pattern 1.Figure 83 is the local figure that amplifies expression pattern 13.

(14) pattern 14: at employed second pattern of the generation of pattern 13, using diffusion length shown in Figure 34 is that 4 error diffusion matrix and diffusion length shown in Figure 35 are 5 the error diffusion matrix error-diffusion method by 0.9: 0.1 the error diffusion matrix that ratio synthesized (Figure 34 * 0.9+ Figure 35 * 0.1), obtains the pattern 14 as the 3rd pattern.Figure 84 is the local figure that amplifies expression pattern 14.

(15) pattern 15: the Monte Carlo method to pattern 14 repeated application 60 times obtains the pattern 15 as the 4th pattern.Figure 85 is the local figure that amplifies expression pattern 15.

The spatial frequency distribution of first pattern A～E is shown in Figure 86, the spatial frequency distribution of pattern 1～15 is shown in Figure 87～91.In addition, Figure 92 is the reduction degree of the low spatial frequency composition that difference realized of the method for making of table of induction diagrammatic sketch case.Shown in Figure 92, as can be known, under the situation of using different arbitrary first pattern in equalization point footpath, also, reduce the low spatial frequency composition efficiently by the application of Hi-pass filter, the further application of error-diffusion method, Monte Carlo method.Particularly, at the 3rd pattern of application error diffusion method and further use in the 4th pattern of Monte Carlo method, the reduction effect of low spatial frequency composition is remarkable.

Under the situation of using Hi-pass filter, different with bandpass filter, do not establish higher limit in the spatial frequency zone of being extracted, therefore also isolated point may take place, but, shown in above-mentioned pattern 1～15, be under the situation of configuration pattern a bit randomly at first pattern that uses, shown in Figure 93, do not find the pilosity of isolated point.

On the other hand, under the situation of using first pattern that disposes light brightness distribution randomly shown in Figure 94, at the pattern of this first pattern being used Hi-pass filter and passing threshold method institute binaryzation, and use Hi-pass filter and by in the pattern of error-diffusion method institute binaryzation, be difficult to isolated point is reduced to sufficient degree, preferably carry out the reduction processing of isolated point by the application of Monte Carlo method.

Figure 95 carries out the binaryzation of the application of Hi-pass filter and threshold method with the same method of the generation of above-mentioned pattern 1 to first pattern shown in Figure 94 with using and the local figure that represents that amplifies of pattern that obtains.Figure 96 carries out the binaryzation of the application of Hi-pass filter and error-diffusion method with the same method of the generation of above-mentioned pattern 2 to first pattern shown in Figure 94 with using and the local figure that represents that amplifies of pattern that obtains.Figure 97 carries out the application of the binaryzation of application, error-diffusion method of Hi-pass filter and Monte Carlo method with using with the same method of the generation of above-mentioned pattern 3 and the local figure that represents that amplifies of pattern that obtains to first pattern shown in Figure 94.Figure 98 is the figure of the isolated point generation number of the pattern shown in expression Figure 95～97.In addition, Figure 99 is the figure that the spatial frequency distribution with the pattern shown in Figure 94～97 compares.Shown in Figure 98 and Figure 99, as can be known, even comprise at first pattern under the situation of high spatial frequency composition morely,, also can obtain the good pattern that the low spatial frequency composition fully reduces and the generation number of isolated point is few by the application of Hi-pass filter and Monte Carlo method.

Claims (10)

1. the generation method of a random pattern possesses:

To disposing first pattern a plurality of points or that dispose light brightness distribution randomly, the wave filter of the low spatial frequency composition of the not enough particular value of spatial frequency is removed or is reduced in application at least from the contained spatial frequency composition of first pattern, generate the operation of second pattern;

By described second pattern is used the tuning method of shivering, generate the operation of the 3rd pattern of the information that is transformed to discretize.

2. the method for claim 1, wherein described the 3rd pattern is the pattern that is transformed to by the information of secondary discretize.

3. method as claimed in claim 2 wherein, also possesses: to being transformed to the 3rd pattern by the information of secondary discretize, by Monte Carlo method isolated black or white pixel are moved, generate the operation of the 4th pattern.

4. the method for claim 1, wherein described tuning method of shivering is an error-diffusion method.

5. method as claimed in claim 4 wherein, by being applied in the error-diffusion method that the scope below 6 pixels spreads mapping fault more than 3 pixels, generates the 3rd pattern.

6. the method for claim 1, wherein described wave filter is: the Hi-pass filter of only removing or reduce the low spatial frequency composition of the not enough particular value of spatial frequency from the contained spatial frequency composition of described first pattern.

7. method as claimed in claim 6, wherein, described wave filter is: only remove from the contained spatial frequency composition of described first pattern or reduce spatial frequency less than 0.01 μ m ^-1The Hi-pass filter of low spatial frequency composition.

8. the method for claim 1, wherein, described wave filter is a bandpass filter, in the described bandpass filter, by from the contained spatial frequency composition of described first pattern, removing or reducing the low spatial frequency composition of the not enough particular value of spatial frequency and remove or reduce the high spatial frequency composition that spatial frequency surpasses particular value, extract the spatial frequency composition of particular range.

9. method as claimed in claim 8, wherein, in the operation of described generation second pattern, the lower limit B of the spatial frequency in the spatial frequency composition of the described particular range that extracts by the application of described bandpass filter is 0.01 μ m ^-1More than and higher limit T be 1/ (D * 2) μ m ^{- 1}Below, at this, D (μ m) is based on printing equipment that the described the 3rd or the 4th pattern prints or based on the resolution of the device of the described the 3rd or the 4th pattern processing concaveconvex shape.

10. method as claimed in claim 9, wherein, by the inverse of the promptly maximum cycle length 1/B of inverse of the lower limit B of the spatial frequency in the spatial frequency composition of described particular range and higher limit T be minimum cycle length 1/T by

BandWidth(％)＝100×(1/B-1/T)/(1/B+1/T)

Represented BandWidth satisfies:

15≤BandWidth(％)≤70。

Images