[go: up one dir, main page]

WO1996024033A1 - Systeme de prediction de proportions et procede de realisation de melanges - Google Patents

Systeme de prediction de proportions et procede de realisation de melanges Download PDF

Info

Publication number
WO1996024033A1
WO1996024033A1 PCT/US1995/000972 US9500972W WO9624033A1 WO 1996024033 A1 WO1996024033 A1 WO 1996024033A1 US 9500972 W US9500972 W US 9500972W WO 9624033 A1 WO9624033 A1 WO 9624033A1
Authority
WO
WIPO (PCT)
Prior art keywords
proportion
characteristic
elements
target
mixture
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/US1995/000972
Other languages
English (en)
Inventor
Hideyuki Takagi
Eiji Mizutani
David M. Auslander
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Kansai Paint Co Ltd
Panasonic Holdings Corp
University of California Berkeley
University of California San Diego UCSD
Original Assignee
Kansai Paint Co Ltd
Matsushita Electric Industrial Co Ltd
University of California Berkeley
University of California San Diego UCSD
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Kansai Paint Co Ltd, Matsushita Electric Industrial Co Ltd, University of California Berkeley, University of California San Diego UCSD filed Critical Kansai Paint Co Ltd
Priority to US08/875,399 priority Critical patent/US6081796A/en
Priority to JP8523482A priority patent/JPH10513290A/ja
Priority to PCT/US1995/000972 priority patent/WO1996024033A1/fr
Publication of WO1996024033A1 publication Critical patent/WO1996024033A1/fr
Priority to MXPA/A/1997/005596A priority patent/MXPA97005596A/xx
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
    • G01J3/46—Measurement of colour; Colour measuring devices, e.g. colorimeters
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
    • G01J3/46—Measurement of colour; Colour measuring devices, e.g. colorimeters
    • G01J3/462—Computing operations in or between colour spaces; Colour management systems
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
    • G01J3/46—Measurement of colour; Colour measuring devices, e.g. colorimeters
    • G01J3/463—Colour matching
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Definitions

  • the present invention relates to a proportion predicting system for predicting a proportion or a mixing ratio at which a plurality of materials, colors, light, acoustic signals, electrical signals, electromagnetic waves or the like are mixed to produce a desired target, and also to a mixture making method for making the desired target by utilization of the proportion predicted by the proportion predicting system.
  • a further prior art method is disclosed by, for example, G. Wyszecki and W.S. Stiles in their "Color Science: Concepts and Methods, Quantitative Data and Formula” (New York, NY: Wiley, 2nd ed. (1982)) , in which a model mathematically simulating the relationship between the color spectrum and the projected image of the propor ⁇ tion, such as represented by the Kubelka-Munk theory is utilized to determine the proportion from the color spectrum.
  • a more recent prior art method is disclosed by, for example, J.M. Bishop, M.J. Bushnell and S. Westland in their "Application of Neural Networks to Computer Recipe Prediction” (Color Research and Application, Vol. 16, No. 1, pp.3-9, (1991) , in which the neural network is taught to output the proportion in response to input of the color spectrum.
  • the element pigments are mixed together according to the proportion so obtained to produce the target color.
  • the utilization of the neural network appears to be an effective technique to remove the difficulty in configuring the substitute model.
  • the human vision is too sensitive to allow the neural network to exhibit a required accuracy in predicting small proportions as low as 0.01%.
  • DISCLOSURE OF THE INVENTION the present invention is intended to provide an improved proportion predicting system for systematically predicting, with high accuracy, proportion or a mixing ratio at which a plurality of element colors, element light, element sound signals, element electrical signals or element electromagnetic waves are mixed to produce a target material, a target color, target light, a target sound signal, a target electrical signal or a target electromagnetic wave.
  • the present invention is also intended to provide an improved mixture making method for making the desired target, such as material, color, light, sound signal, electrical signal or electromagnetic wave, having a specification similar to that of the target material, target color, target light, target sound signal, target electrical signal or target electromagnetic wave.
  • desired target such as material, color, light, sound signal, electrical signal or electromagnetic wave
  • the term "element” herein used is to be understood as meaning a material, color, light, an acoustic signal, an electric signal or an electromagnetic wave having a variety of physical and/or chemical characteristics having a capability of being mixed and that the term “target” also herein used is to be understood as meaning a material, color, light, an acoustic signal, an electric signal or an electromagnetic wave produced by mixing these elements.
  • the term “mixing” herein recited is to be understood as meaning not only a direct mixing of physical matter, chemical matter or electric matter, but also a mixing of matter which human sensory organs so perceive even though no direct mixing of matter takes place.
  • the mixing of matter which human sensory organs so perceive includes preparation of a mixed color by spatially disposing colored toner particles in a given area at varying proportions by means of, for example, a color printer, or by displaying by means of, for example, a color display, colors at varying intensities of light spatially distributed in the color display.
  • mixture herein used is to be understood as meaning material, color, light, acoustic signal, electric signal or electromagnetic wave so prepared by the specific mixing.
  • a proportion predicting apparatus which includes a proportion characteristic extractor for deter- mining a characteristic of a target, a GA (genetic algo ⁇ rithm) calculator for predicting the proportion or mixing ratio of two or more elements, and an evaluating means for calculating a similarity between two characteristics.
  • a mixture making method which includes a proportion characteristic extractor for deter ⁇ mining a characteristic of a target, a GA (genetic algo ⁇ rithm) calculator for predicting the proportion or mixing ratio of two or more elements, an evaluating means for calculating a similarity between two characteristics and a mixer for mixing the elements together at respective proportions finally calculated by the GA processor.
  • a proportion characteristic extractor for deter ⁇ mining a characteristic of a target
  • a GA genetic algo ⁇ rithm
  • the evaluating means employed in the present invention includes at least one of a mixture element evaluating section adapted to receive the proportions predicted by the GA processor to evaluate the type of the elements of which proportion is not zero, and/or a mixture characteristic evaluating section for predicting a charac ⁇ teristic of the mixture formed by mixing the elements at the respective proportions and for comparing it with the characteristic of the target.
  • the mixture element evaluating section of the evaluating means used in the present invention includes at least one of an element number evaluating section and an unnecessary element number evaluating section.
  • the element number evaluating section includes a mixture element number selector adapted to receive the proportions predicted by the GA processor to digitalize the predicated proportions, a mixture element predictor adapted to receive the charac- teristic extracted by the mixture characteristic extractor to predict the type of elements that form the target, and an element number distance calculator for comparing the element number, obtained from the mixture element number selector for use in mixing, with the element number obtained from the mixture element predictor and for outputting a similarity.
  • the unneces ⁇ sary element number evaluating section includes a mixture element number selector, a knowledge base in which unneces ⁇ sary combinations for the mixing are described in the form of knowledge, and a penalty section adapted to receive the element number obtained from the mixture element number selector for use in mixing and for making reference to the knowledge base to lower a fitness in the event of the presence of the unnecessary element number.
  • the mixture characteristic evaluating section includes a mixture characteristic predictor adapted to receive the predicted proportions calculated by the GA processor and to predict the characteristic of the mixture obtained by mixing the elements at the predicted propor- tions, and a mixture characteristic distance calculator for comparing the characteristic extracted by the mixture characteristic extractor, with the characteristic of the mixture predicted by the mixture characteristic predictor and for outputting a similarity.
  • the GA processor used in the present invention includes a GA initial value determining section for determining initial values for predicting the proportions according to the genetic algorithm, and a dynamic GA processor for sequentially determining the proportions.
  • the GA initial value determining section includes at least an initial proportion value determining section adapted to receive the characteristic of the target and to predict the proportions which are then rendered to be initial values.
  • the GA initial value determining section referred to above may include at least an initial proportion value determining means for receiving the characteristic of the target to predict the proportions which are then rendered to be respective initial values, a knowledge base concern- ing combinations of the elements, and a multi-elite generator for comparing an output from the initial propor ⁇ tion value determining means with the knowledge to formu ⁇ late new initial values of the GA processor by replacing proportions of unnecessary element candidates with zero.
  • the GA initial value determining section may includes a color space characteristic extractor for outputting coordinates in a color space such as L*-a*- b* of the color of the target, a color space classifier adapted to receive the coordinates in the color space obtained by the color space characteristic extractor to determine which representative color area the color of the target is included, a knowledge base in which knowledge concerning combinations of colors of the elements are described, a random initial value generator for outputting random proportions, and a random initial value corrector adapted to receive the proportion generated by the random initial value generator and information on the representa ⁇ tive color area determined by the color space classifier and operable to make a search in the knowledge base to determine if an improper color combination in the color area is included in a random value of the proportion obtained from the random initial value and to correct the improper random proportion.
  • a color space characteristic extractor for outputting coordinates in a color space such as L*-a*- b* of the color of the target
  • a color space classifier adapted to receive the coordinates in the color
  • the characteristic of the given target is first determined by the mixture characteristic extractor.
  • a characteristic may be, in the case of sound or color, a physical characteristic such as a spectrum.
  • the GA processor then predicts, on the basis of the genetic algorithm, the mixing ratio of the mixture of a specification formulated by mixing two or more elements.
  • the evaluating means compares the characteristic determined by the mixture characteristic extractor with the characteristic of the mixture which may eventually prepared by mixing the elements at the respective proportions predicted by the Ga calculator and also evaluates the degree of accuracy of the proportions predicted by the GA processor.
  • the proportion predictor is comprised of the above described mixture characteristic extractor, the GA processor and the evaluating means.
  • the mixer serves to mix the elements at the respective predicted proportions to provide the mixture.
  • the GA processor utilizes as the respective initial values the proportions determined by the initial value determining section so that the accuracy of predic ⁇ tion of the mixing ratio can be progressively increased on the basis of the genetic algorithm.
  • the evaluating means output a result of evaluation obtained by the built- in element number evaluating section, the unnecessary element number evaluating section or the mixture character ⁇ istic evaluating section.
  • the element number evaluating section is comprised of the mixture element number selec ⁇ tor, the mixture element predictor and the element number distance calculator.
  • the mixture element number selector is operable, in response to the proportions predicated by the GA processor, to digitalize the predicted mixing ratio, that is, to render the predicted proportions to be OFF if they are lower than a predetermined threshold value, but ON if they are higher than the predetermined threshold.
  • the mixture element predictor receives the characteristic extracted by the mixture characteristic extractor so that the type of the elements forming the target can be predicted.
  • the element number distance calculator compares the element number, which is obtained from the mixture element number selector and which is used in mixing, with the element number obtained by the mixture element predictor and them outputs a similarity between them.
  • the unnecessary element number evaluating section is comprised of the mixture element number selector, the knowledge base and the penalty section.
  • the knowledge base has knowledge necessary to eliminate redundant elements
  • the penalty section receives the element number obtained from the mixture element number selector for use in mixing and refers to the knowledge base so that the fitness can be lowered in the event of the presence of the unnecessary element combinations.
  • the mixture characteristic evaluating section is comprised of the mixture characteristic predictor and the mixture characteristic distance calculator.
  • the mixture characteristic predictor receives the proportions predicted by the GA processor and predicts the characteristic of the mixture which is obtained by mixing the elements at the respective predicted proportions.
  • the mixture characteris ⁇ tic distance calculator compares the characteristic obtained by the mixture characteristic extractor with the characteristic of the mixture obtained by the mixture characteristic predictor and then provides an output indicative of similarity between them.
  • Fig. 1 is a block diagram showing a proportion predicting apparatus according to the present invention
  • Fig. 2 is a block diagram showing an output of a GA processor used in the apparatus of the present inven ⁇ tion;
  • Fig. 3 is a flowchart showing a genetic algorithm employed in the practice of the present invention;
  • Fig. 4 is a block diagram showing a method of making a mixture according to the present invention
  • Fig. 5 is a block diagram showing the details of an evaluating means used in the apparatus of the present invention
  • Fig. 6 is a block diagram showing the details of a mixture element predictor used in the apparatus of the present invention.
  • Fig. 7 is a block diagram showing the details of a mixture characteristic predictor used in the apparatus of the present invention
  • Fig. 8 is a block diagram showing the details of the Ga calculator used in the apparatus of the present invention.
  • Fig. 9 is an explanatory diagram showing a a*-b* color space and a fuzzy-classification of the a*-b* color space.
  • the present invention will be described as applied to colorant proportion prediction, that is, prediction of the mixing ratio of colorants to be mixed to produce a desired color, as is the case with the prior art hereinbefore discussed.
  • Fig. 1 there is shown a schematic block representation of a proportion predicting system according to one preferred embodiment of the present invention.
  • the proportion predicting system shown therein comprises a mixture characteristic extractor 1, an evaluat ⁇ ing unit 2 and a GA (genetic algorithm) processor 3.
  • GA genetic algorithm
  • the mixture characteristic extractor 1 analyzes and output a color spectrum of the target color.
  • the spectrum represents physical properties continuously appearing on the axis of frequency, in most cases the spectrum is expressed in terms of n values using n filter banks and n discrete Fourier transforms.
  • the GA processor 3 makes use of a genetic algorithm (GA) to predicate how to mix m element pigments (m being greater than 1) in a particular mixing ratio to produce the target color.
  • Fig. 2 illus- trates examples of mixing ratios or proportions of the element pigments which have been predicted by the GA processor 3, and those mixing ratios are outputted from the GA processor 3.
  • GA genetic algorithm
  • the evaluating unit 2 compares information on the target color fed from the mixture characteristic extractor 1 with the predicted proportion fed from the GA processor 3 to determine, and then outputs a fitness (or an evaluated value) descriptive of, the degree of accuracy of the predicted proportion, that is, the degree of color match between the target color and the color represented by the predicted proportion.
  • the GA processor 3 makes use of the output value from the evaluating unit 2 to progressively increase the degree of accuracy of the predicted proportion according to the genetic algorithm. Referring to Figs. 2 and 3, the GA processor 3 operates according to the genetic algorithm, and proportion vectors to be predicted by the GA processor 3 are treated as n chromosomes of the genetic algorithm such as shown in Fig. 2.
  • the proportion vectors to be predicted represent an -order vector.
  • the proportion vector 1 shown in Fig. 2 it means that m element pigments, i.e., white, green-1, green-2, yellow and red pigments are mixed in a proportion of 0.23, 0.04, 0.31, 0 and 0.11, respectively.
  • the GA processor 3 deals with the correspondingly plural proportion vectors and successively increases their accuracy of prediction.
  • Fig. 3 illustrates a flowchart showing how the accuracy of prediction of the proportions predicted by the GA processor 3 is increased.
  • the flowchart shown in Fig .3 is based on the genetic algorithm.
  • the n chromosomes that is, the propor ⁇ tions, are first initialized.
  • the simplest conventional initializing method is to initialize them with random values.
  • the initializing method is uniquely designed as will be described later.
  • the proportions so initialized are supplied to the evaluating unit 2.
  • the accuracy of prediction of the proportions done at the first cycle may not be high and, therefore, the evaluating unit 2 may not give a high fitness to the n proportion vectors.
  • the fitnesses are again fed back to the GA processor 3 which then select a number of predicted proportion vectors which were given a high fitness according to the fitnesses so fed back to the Ga processor 3.
  • the propor- tion vectors selected by the GA processor 3 represent parents that produce offspring proportion vectors of the succeeding generation. According to the genetic algorithm this process is referred to as selection.
  • the next step is to cross over two predicted proportion vectors arbitrarily chosen from the selected parent proportion vectors to thereby provide predicted proportion vectors of the next succeeding generation. This process is carried out by crossover operation to combine the two m-order proportion vectors to produce different two m-order proportion vectors. The simplest one-point crossover will now be discussed as an example.
  • the two m-order proportion vectors are represented by chromosomes a and b.
  • the one- point crossover is a calculation to produce first offspring proportion vectors from an r-order element of the first half of the chromosome a and a (m-r)-order element of the latter half of the chromosome b and to produce second offspring proportion vectors from an r-order element of the first half of the chromosome b and a (m-r)-order element of the latter half o the chromosome a, wherein 0 ⁇ r ⁇ m.
  • This crossover is referred to as the one-point crossover since offsprings are produced by changing the elements at a boundary between the r-order and the next succeeding (r+1) order from the beginning.
  • n proportion vectors of the second generation are produced.
  • Calculation performed at the end of the genetic algorithm is referred to as mutation in which random values are added to values descriptive of the proportions of arbitrarily chosen element of the arbitrarily chosen chromosomes.
  • this mutation corresponds to a general search for preventing the propor ⁇ tions from falling into local minimum.
  • the n chromosomes obtained by the foregoing process of selection, crossover and mutation represent the proportion vectors of the second generation.
  • FIG. 4 there is shown a method of preparing a mixture according to one preferred embodi ⁇ ment of the present invention.
  • reference numeral 4 represents a proportion predicator and reference numeral 5 represents a mixer.
  • the proportion predicator 4 is comprised of the mixture characteristic extractor 1, the evaluating unit 2 and the GA processor 3 all shown in and described with reference to Fig. 1 and has a function identical with that of the proportion predicting system shown in Fig. 1.
  • the mixer 5 operates to mix element pigments at the proportion determined by the proportion predicator 4. Functions of various components will be described in further detail hereinafter.
  • Fig. 5 illustrates one embodiment of the evaluat ⁇ ing unit 2 employed in the practice of the present inven ⁇ tion.
  • the evaluating unit 2 comprises a mixture element evaluating unit 6, a mixture characteristic evaluating unit 7, and a fitness integrator 8.
  • the mixture element evaluating unit 7 shown therein performs evaluation from two viewpoints.
  • This mixture element evaluating unit 6 includes an element number evaluating section 61, and an element number unnecessity evaluating section 63.
  • the mixture element evaluating unit 6 evaluates what element pigments are to be mixed together. In other words, this mixture element evaluating unit 6 serves to convert the proportion predicted by the GA processor 3 into one of binary information (ON or OFF) indicative of whether the element pigments are to be mixed or whether they are not to be mixed and compare it with physical characteristics of the target color.
  • a circuit structure including the element number evaluating section 61 and the element number unnecessity evaluating section 63 performs a specific evaluation.
  • the mixture characteristic evaluating unit 7 evaluates the value of the proportion predicted by the GA processor 3.
  • the fitness integrator 8 serves to apply a weight on each of a plurality of fitness- es and then to sum them together so that one fitness can be outputted therefrom as an output from the evaluating unit 2.
  • the simplest weighing method is to apply an equal weight.
  • the element number evaluating section 61 shown in Fig. 5 includes a mixture element number selector 620, a mixture element predicator 611 and an element number distance calculator 612.
  • the element number evaluating section 61 of the construction described above functions in the following manner.
  • the mixture element number selector 620 receives the proportions, such as shown in Fig. 2, which have been predicted by the GA processor 3 and converts them into one of binary digits consisting of OFF information, representing that the element pigments at a proportion lower than a threshold value are not mixed, and ON information representing that the element pigments at any other proportion are mixed together.
  • the threshold value referred to above is determined in consideration of the visual perceptibility of whether one can perceive a difference when the element pigments are mixed, and the accuracy of prediction of the GA processor 3.
  • the mixture element predicator 611 makes use of a color spectrum obtained from the mixture characteristic extractor 1 to directly predicate which element pigments are to be mixed together.
  • Fig. 6 illustrates an example in which the mixture element predicator 611 is implemented by a neural network.
  • reference numeral 6111 represents a characteristic-element number converter
  • reference numeral 6112 represents a threshold processor.
  • the characteristic-element number converter 6111 is in the form of a neural network of a three-layered fed-forward type and is adapted to receive the n-order color spectrum obtained by the mixture characteristic extractor 1 so that either an ON output for mixing the m-order element pigments or an OFF output for not mixing the m-order element pigments can be outputted therefrom.
  • the neural network for the character ⁇ istic-element number converter 6111 acquires knowledge from a ready-made training data set.
  • the element pigments are beforehand mixed together to provide a paint of a represen ⁇ tative color which is subsequently subjected to colorimetr- ic measurement to determine the color spectrum of the resultant paint.
  • an accurate relationship between the color spectrum and the m-order ON or OFF information can be obtained. This information is used as a teacher's signal in allowing the neural network to acquire knowledge.
  • the characteristic- element number converter 6111 can easily be implemented. In practice, however, even if the neural network is taught to output one of binary digits, 0 or 1, it will not output a complete 0 or 1 and rather 0.001 or 0.998 since an output layer of the neural network make use of a continuous function such as a sigmoid function.
  • the threshold processor 6112 serves to digitalize these values.
  • the threshold processor 6112 compares the characteristic- element number converter 6111 with a predetermined thresh ⁇ old set in it to forcibly convert the output of the neural network into one of the ON and OFF signals.
  • the ON and OFF signals are outputted from the threshold processor 6112 if the value is greater than 0.5 and smaller than 0.5, respectively.
  • the element number distance calculator 612 operates to make a comparison between the information of which element pigments are to be mixed in the light of the color spectrum of the target paint obtained from the mixture element predicator 611 and the information of which element pigments are to be mixed in the light of the predicted proportion obtained from the mixture element number selector 620 and then to provide an output indica ⁇ tive of the degree of difference between these pieces of information.
  • the element number distance calculator 612 determines two m-order binary vector distances. The simplest method of accomplishing this is to calculate the number of bits that are different. This distance is then fed to the fitness integrator 8.
  • Another specific example of the mixture element evaluating unit 6 is the element number unnecessity evaluating section 63.
  • the element number unnecessity evaluating section 63 shown in Fig. 5 includes the mixture element number selector 620, a knowledge base 631 and a penalty calculator 632.
  • the element number unnecessity evaluating section 63 of the above described construction operates in the following manner.
  • the penalty calculator 632 is adapted to receive from the mixture element number selector 620 information on which element pigments to be mixed together and makes access to the knowledge base 631 to determine if a combina ⁇ tion of the element pigments inputted from the mixture element number selector 620 is unacceptable or unnatural.
  • the knowledge base 631 stores therein knowledge associated with the color combination such as, for example:
  • Rule 1 Avoid use of complementary color pigments, e.g., red and green pigments
  • Rule 2 Avoid use of the same color pigments, e.g., green-1 and green-2 pigments
  • Rule 3 Keep the proportion about 100%.
  • the penalty calculator 632 generates evaluated values depending on to what extent the predicted proportion obeys these rules.
  • this mixture characteristic evaluating unit 7 includes a mixture characteristic distance predictor 72 which operates to interpret, on the basis of the proportion predicted by the GA processor 3, perceived attributes of the predicted color, e.g. , coordinates of the L*-a*-b* values in the CIE 1976 (x,y)-chromaticity diagram. It is to be noted that L*, a* and b* represent lightness, hue and chroma.
  • This mixture characteristic distance calculator 71 receives the target coordinates of the L*-a*-b* from the mixture characteristic extractor 1 and calculates the color distance, Euclidean distance, between the two m-order proportion vectors in the L*-a*-b* space.
  • the L*, a* and b* values are calculated from the surface spectral reflect- ance.
  • the element pigments are beforehand mixed together to provide a paint of a represen ⁇ tative color which is subsequently subjected to colorimetr- ic measurement to determine the color spectrum of the resultant paint.
  • the relationship between the proportion of the element pigments and the color space coordinates is available beforehand and this is used as a training data for the neural network of a mixture charac ⁇ teristic predicator 72.
  • the color of the paint obtained when the element pigments have been mixed according to the proportion predicted by the GA processor 3 and the color of the target paint are converted into the distance in the color space which is subsequently fed to the fitness integrator 8.
  • the evaluating unit 2 evaluates the accuracy of the proportion predicted by the GA processor 3.
  • the evaluation is fed back to the GA processor 3.
  • the evaluating unit 2 has three compo ⁇ nents, i.e., the element number evaluating section 61, the element number unnecessity evaluating section 63 and the mixture characteristic evaluating unit 7 to perform evaluation, the use of only one of them is sufficient for the evaluating unit 2 to perform evaluation.
  • the use of the three components for the evaluating unit 2 is advantageous in that a higher accuracy can be expected, but the use of these components in combination is not always essential. (See the Table below.) When a paint is prepared by mixing the element pigments, no one can see what color it will present unless they are actually mixed together.
  • Fig. 8 illustrates the structure of the GA processor 3 according to one preferred embodiment of the present invention.
  • the GA processor 3 includes an initial value determining unit 31 and a dynamic GA determining unit 32.
  • the GA processor 3 operates in the following manner according to the genetic algorithm. Since the genetic algorithm is widely employed in the art, terms “population size”, “chromosome”, “gene” and so on which are used in association with the genetic algorithm are herein used with no definition given as they are believed to be well understood by those skilled in the art. Chromosomes outputted from the GA processor 3 are shown in the diagram. Proportions, that is, the mixing ratios, of the m element pigments combined in the chromo ⁇ somes correspond respectively to m genes.
  • n such chromosomes are prepared, wherein n represents the population size.
  • the genes of each of the n chromosomes are initialized by the initial value determining unit 31, causing the GA processor 3 to output an initial output.
  • the initial value determining unit 31 operates only one time at the outset.
  • Values of the chromosomes evaluated by the evaluating unit 2 are returned to the GA processor 3.
  • the dynamic GA determining unit 32 performs, on the basis of the fitnesses, a genetic operations (selection, crossover and mutation) according to the genetic algorithm to produce n chromosomes of the next generation.
  • the chromosomes that is, predicted propor ⁇ tions
  • the dynamic GA determining unit 32 repeats this procedure.
  • the dynamic GA determining unit 32 plays an important role of the GA processor 3.
  • the initial value determining unit 31 includes a counter 311, a control unit 312, an initial proportion determiner 313, a multi-elite generator 314, a random initial value generator 315, a color space characteristic extractor 316, a color space classifier 317 and a random initial value corrector 318.
  • the initial value determining unit 31 of the construction described above carries out an efficient initialization using three methods such as hitherto carried out in the genetic algorithm.
  • the first initialization is carried out in the initial proportion determiner 313 which is adapted to receive from the mixture characteristic extractor 1 information on the color spectrum of the target paint and to output the proportion, that is, the mixing ratio, of the element pigments.
  • the initial proportion determiner 313 itself may be said to be a proportion predicting apparatus and can be implemented by a neural network. Such a neural network is reported in the Bishop et al. paper hereinbefore discussed in connection with the prior art and can be implemented in a fashion similar to the neural network shown in any one of Figs. 6 and 7. As shown by experimen- tal data as will be discussed later, the initial proportion determiner 313 alone does not provide a sufficient predic ⁇ tion.
  • the output from the initial proportion value determining unit 313 may be expected to be close to the proportion to be eventually predicted by the proportion predicting apparatus as compared with the random value used for initialization during execution of the standard genetic algorithm. Accordingly, if the predicted proportion obtained from the initial proportion determiner 313 is used in the proportion predicting system of the present inven ⁇ tion as the initial value, the predicting performance can be increased.
  • the second initialization is carried out in the initial proportion determiner 313, the knowledge base 631 and the multi-elite generator 314.
  • the initial proportion determiner 313 With the initial proportion determiner 313, a single initial value of a good birth can be obtained.
  • the initial proportion determiner 313 fails to take into consideration a relation ⁇ ship between the complementary color and the same color and the predicted proportion descriptive of how and what element pigments are to be mixed often contains unnecessary and/or useless combinations of the element pigments.
  • the multi-elite generator 314 is adapted to receive the predicted proportion from the initial proportion determiner 313 and outputs a predicted proportion of a good birth -30-
  • the multi-elite generator 314 serves to provide better starting points for genetic search.
  • the third initialization is carried out in the random initial value generator 315, the color space characteristic extractor 316, the color space classifier 317, the random initial value corrector 318 and the knowledge base 631.
  • the random initial value generator 315 randomly outputs proportions in a manner similar to the random initial value generator used for initialization of the standard genetic algorithm.
  • the color space character ⁇ istic extractor 316 performs calculation to determine the coordinates of the color space such as the L*-a*-b* space from the spectrum of the target color.
  • the color space classifier 317 divides or classifies the color space into representative color areas and provides an output descrip ⁇ tive of the extent to which the coordinates of the color space obtained from the color space characteristic extrac- tor 316 belongs to what representative color area divided by such classifier 317.
  • FIG. 9 illustrates an example in which the two dimensional a*-b* space is projected into the hue angle tan 1 (b*/a*) and the hue angle is fuzzy-classified into five representative color areas, i.e., red (R) , yellow (Y) , green (G) , blue (B) and violet (V) color areas.
  • Triangles in Fig. 9 define respective degrees of belonging in correspondence with the hue angles of red (R) to violet (V) . The color varies continuously and the degree of belonging progressively shifts from a certain representa- tive color area to the next one with change in hue angle.
  • FIG. 9 also illustrates an output generated from the color space classifier 317 when one color space coordinate (a*., b*i) is obtained from the color space characteristic extractor 316.
  • the color space classifier 317 provides an output indicating that, in the case of the hue angle tan 1 (b* i /a* 1 ) for this coordinates, 0.75 and 0.25 belong to the red and yellow color areas, respectively, but not to any other color areas.
  • the random initial value corrector 318 receives this degree of belonging and also the random initial value for the proportion of the element pigments from the random initial value generator 315.
  • the random initial value corrector 318 refers to the degree of belonging to the representative color obtained from the color space classifier 317 and the knowledge of the color combination obtained from the knowledge base 631 in order to modify the randomly generated proportions.
  • the random initial value corrector 318 forcibly renders green proportions to be zero because of the red-green complementary color relationship, and renders to zero value either red-1 or red-2 proportion.
  • the foregoing three initializations are con ⁇ trolled by the counter 311 and the control unit 312.
  • the counter 311 is initialized to zero.
  • the control unit 312 operates only when the count of the counter 311 reads zero.
  • the population is initialized by the initial proportion determiner 313, the multi-elite genera ⁇ tor 314 and the random initial value corrector 318. Since the random initial value generator 315 can initialize any number, some are initialized at the outset by the initial proportion determiner 313 and then the outputs from the initial proportion determiner 313 are modified by the multi-elite generator 314 according to some rules in the knowledge base 631 while the remainder are initialized by the random initial value corrector 318.
  • the initial population so formed in the manner described above is fed to the dynamic GA determining unit 32.
  • the output from the GA processor 3 is returned to the evaluating unit 2 and the value evaluated by the evaluating unit 2 is fed back to the GA processor 3, the count of the counter 311 is incremented by one. Since the count of the counter 311 is no longer zero, initialization of the initial value determining unit 31 takes place only one time at the beginning.
  • each method can be expected to give rise to the value of a good birth as compared with the initializa ⁇ tion with the random value such as executed in the prior art genetic algorithm. Accordingly, even though the three initialization methods are not realized simultaneously, it is expected that the predicting performance and the manufacturing performance are high as compared with the prior art proportion predicting system and the mixture making method.
  • color space classifier 317 used in the embodiment of the present invention has been shown as operable to fuzzy-classify the color space, similar effects can be obtained even if a crisp division is carried out, since there is a significance in that the initial value of a good birth is determined with the use of the knowledge of each divided color area.
  • the present invention has been described as applied to production of color paint generally matching with the color of the target paint prepared by mixing the element pigments, the present invention should not be limited to production of such a physical mixture and can be equally applicable to mixing of waves such as color, light and sound. Also, although the disclosed embodiments are directed to physical mixtures, the present invention is also applicable to mixture of any matter the human being can percept. By way of example, in the case of color formulation based on visually perceptible mixture such as a color printer, the proportion predicting apparatus of the present invention may operate to predicate the proportion of surface areas of element colors spatially distributed.
  • the proportion predicting apparatus of the present invention may operate to predicate the mixing ratio of light intensities of color elements distributed spatially.
  • the mixture making method of the present invention may have a number of applications.
  • the proportion at which the elements are to be mixed to produce the target of a predetermined specification can be predicted with high accuracy, with no need to make an actual mixture of elements. Consequently, a mixture substantially matching with the target of the predetermined specification can be easily prepared. This brings about such advantages that the length of time required to perform the prediction and the manufacture and the associated cost can be reduced.
  • the present invention has a number of industrial utilities.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Physiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Genetics & Genomics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Spectrometry And Color Measurement (AREA)

Abstract

L'invention a pour objet un système de prédiction de proportions pour obtenir une cible déterminée par mélange d'un nombre prédéterminé d'éléments selon une proportion déterminée. Ce système comprend un extracteur (1) de caractéristiques cibles pour extraire une caractéristique de la cible prédéterminée, un évaluateur (2) pouvant fonctionner lors de la réception de vecteurs de proportions, lesquels sont représentés par la caractéristique de la cible prédéterminée et la quantité de chacun des éléments, respectivement, pour déterminer l'adéquation des vecteurs de proportions en fonction de la caractéristique extraite de la cible prédéterminée; et un processeur d'algorithme génétique (3) pour prédire les vecteurs de proportion en fonction de l'adéquation selon un algorithme génétique, dans lequel la quantité de chacun des éléments et chacun des vecteurs de proportion sont représentés par un gène et un chromosome, respectivement. Les vecteurs de proportion, qui ont été prévus par le processeur d'algorithme génétique, sont entrés dans l'évaluateur pour amener ce dernier à répéter l'évaluation des vecteurs de proportion pour déterminer les vecteurs de proportion optimaux.
PCT/US1995/000972 1995-01-31 1995-01-31 Systeme de prediction de proportions et procede de realisation de melanges Ceased WO1996024033A1 (fr)

Priority Applications (4)

Application Number Priority Date Filing Date Title
US08/875,399 US6081796A (en) 1995-01-31 1995-01-31 Proportion predicting system and method of making mixture
JP8523482A JPH10513290A (ja) 1995-01-31 1995-01-31 比率予測システムと混合物生成方法
PCT/US1995/000972 WO1996024033A1 (fr) 1995-01-31 1995-01-31 Systeme de prediction de proportions et procede de realisation de melanges
MXPA/A/1997/005596A MXPA97005596A (en) 1997-07-23 Prediction system of proportions and method to obtain a mez

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/US1995/000972 WO1996024033A1 (fr) 1995-01-31 1995-01-31 Systeme de prediction de proportions et procede de realisation de melanges

Publications (1)

Publication Number Publication Date
WO1996024033A1 true WO1996024033A1 (fr) 1996-08-08

Family

ID=22248557

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US1995/000972 Ceased WO1996024033A1 (fr) 1995-01-31 1995-01-31 Systeme de prediction de proportions et procede de realisation de melanges

Country Status (2)

Country Link
JP (1) JPH10513290A (fr)
WO (1) WO1996024033A1 (fr)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR2758899A1 (fr) * 1997-01-24 1998-07-31 Hewlett Packard Co Procede de selection de termes d'ordres superieurs pour un reseau neural holographique
WO2000063789A3 (fr) * 1999-04-16 2001-03-22 Siemens Ag Oesterreich Procede permettant de prevoir l'evolution d'une serie de donnees et dispositif de mise en oeuvre dudit procede
EP1146446A1 (fr) * 2000-04-12 2001-10-17 Unilever N.V. Méthode pour optimiser des formulations
EP1146447A1 (fr) * 2000-04-12 2001-10-17 Unilever N.V. Méthode pour trouver des solutions
DE10233022A1 (de) * 2002-07-20 2004-04-15 Zinn, Peter, Dr. Verfahren und Anlage zur Lösung von Aufgaben der adaptiven Chemie
US6757667B1 (en) 2000-04-12 2004-06-29 Unilever Home & Personal Care Usa, Division Of Conopco, Inc. Method for optimizing formulations
US6768973B1 (en) 2000-04-12 2004-07-27 Unilever Home & Personal Care Usa, Division Of Conopco, Inc. Method for finding solutions
US7148900B2 (en) 2000-06-16 2006-12-12 Danichisekia Color And Chemicals Mfg. Co., Ltd. CCM calculating system, CCM calculating method and recording medium
JP2017508952A (ja) * 2013-12-20 2017-03-30 ビーエーエスエフ コーティングス ゲゼルシャフト ミット ベシュレンクテル ハフツングＢＡＳＦ Ｃｏａｔｉｎｇｓ ＧｍｂＨ 色処方を決定するための方法及びシステム
CN116287546A (zh) * 2023-02-08 2023-06-23 东北大学 配料方法、装置、存储介质及计算机设备

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002042105A (ja) * 2000-07-28 2002-02-08 Kurabo Ind Ltd コンピュータカラーマッチング方法
US6795751B2 (en) * 2000-12-06 2004-09-21 Honeywell International Inc. System and method to accomplish high-accuracy mixing
JP7363589B2 (ja) * 2020-03-04 2023-10-18 トヨタ自動車株式会社 塗装品質予測装置および学習済みモデルの生成方法

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5410637A (en) * 1992-06-18 1995-04-25 Color And Appearance Technology, Inc. Color tolerancing system employing fuzzy logic

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5410637A (en) * 1992-06-18 1995-04-25 Color And Appearance Technology, Inc. Color tolerancing system employing fuzzy logic

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
K. CHAN, Computer-Aided Molecular Design Using Genetic Algorithms, PURDUE UNIVERSITY DISSERTATION, May 1994, pages 40-62. *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR2758899A1 (fr) * 1997-01-24 1998-07-31 Hewlett Packard Co Procede de selection de termes d'ordres superieurs pour un reseau neural holographique
WO2000063789A3 (fr) * 1999-04-16 2001-03-22 Siemens Ag Oesterreich Procede permettant de prevoir l'evolution d'une serie de donnees et dispositif de mise en oeuvre dudit procede
EP1146446A1 (fr) * 2000-04-12 2001-10-17 Unilever N.V. Méthode pour optimiser des formulations
EP1146447A1 (fr) * 2000-04-12 2001-10-17 Unilever N.V. Méthode pour trouver des solutions
US6757667B1 (en) 2000-04-12 2004-06-29 Unilever Home & Personal Care Usa, Division Of Conopco, Inc. Method for optimizing formulations
US6768973B1 (en) 2000-04-12 2004-07-27 Unilever Home & Personal Care Usa, Division Of Conopco, Inc. Method for finding solutions
US7148900B2 (en) 2000-06-16 2006-12-12 Danichisekia Color And Chemicals Mfg. Co., Ltd. CCM calculating system, CCM calculating method and recording medium
DE10233022A1 (de) * 2002-07-20 2004-04-15 Zinn, Peter, Dr. Verfahren und Anlage zur Lösung von Aufgaben der adaptiven Chemie
DE10233022B4 (de) * 2002-07-20 2004-09-16 Zinn, Peter, Dr. Verfahren zur Lösung von Aufgaben der adaptiven Chemie
WO2004010347A3 (fr) * 2002-07-20 2004-10-28 Peter Zinn Procede et installation pour resoudre des problemes de chimie adaptative
JP2017508952A (ja) * 2013-12-20 2017-03-30 ビーエーエスエフ コーティングス ゲゼルシャフト ミット ベシュレンクテル ハフツングＢＡＳＦ Ｃｏａｔｉｎｇｓ ＧｍｂＨ 色処方を決定するための方法及びシステム
CN116287546A (zh) * 2023-02-08 2023-06-23 东北大学 配料方法、装置、存储介质及计算机设备

Also Published As

Publication number Publication date
MX9705596A (es) 1997-11-29
JPH10513290A (ja) 1998-12-15

Similar Documents

Publication Publication Date Title
US6081796A (en) Proportion predicting system and method of making mixture
WO1996024033A1 (fr) Systeme de prediction de proportions et procede de realisation de melanges
Puzicha et al. On spatial quantization of color images
CA2452961C (fr) Systemes et methode de choix automatique des caracteristiques visuelles de mise en evidence d'une cible sur un fond
EP0735511A2 (fr) Méthode et appareil pour changer la couleur spécifiée dans une image couleur
AU762994B2 (en) Method for determining printer dye levels
EP0693738A2 (fr) Dispositif et méthode pour produire des masques colorés
US20060022994A1 (en) Method and system of improved color selection
US5012431A (en) Objective color notation system
JPH04227585A (ja) 改良した制御データ列の発生方法及びその装置
JPH07170421A (ja) 色変換方法
Um et al. A study of the emotional evaluation models of color patterns based on the adaptive fuzzy system and the neural network
Neufeld et al. The evolution of artistic filters
CA2211194A1 (fr) Systeme de prediction de proportions et procede de realisation de melanges
Bittermann et al. Visual perception with color for architectural aesthetics
WO1999066430A1 (fr) Procede et appareil pour determiner le motif d'une resine
Pham Spline-based color sequences for univariate, bivariate and trivariate mapping
MXPA97005596A (en) Prediction system of proportions and method to obtain a mez
EP0407381B1 (fr) Systeme et procede de caracterisation objective de couleurs
JPH08116460A (ja) カラーマスキングパラメータ決定装置
JP3109336B2 (ja) 色再現方法
HK1006591B (en) Objective color characterization system and method
Lim Evolving facial expressions
JP3374650B2 (ja) 画像の色分布曲線近似システムおよびこの近似システムを用いた限定色決定システム
Kalra GamOpt: A tool for visualization and optimization of gamuts

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): CA JP MX US

DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
ENP Entry into the national phase

Ref document number: 2211194

Country of ref document: CA

Ref country code: CA

Ref document number: 2211194

Kind code of ref document: A

Format of ref document f/p: F

WWE Wipo information: entry into national phase

Ref document number: PA/a/1997/005596

Country of ref document: MX

ENP Entry into the national phase

Ref country code: JP

Ref document number: 1996 523482

Kind code of ref document: A

Format of ref document f/p: F

WWE Wipo information: entry into national phase

Ref document number: 08875399

Country of ref document: US