EA022480B1 - The method of recognition and identification of patterns and ornaments and intellectual information system for its implementation - Google Patents

The method of recognition and identification of patterns and ornaments and intellectual information system for its implementation Download PDF

Info

Publication number
EA022480B1
EA022480B1 EA201100606A EA201100606A EA022480B1 EA 022480 B1 EA022480 B1 EA 022480B1 EA 201100606 A EA201100606 A EA 201100606A EA 201100606 A EA201100606 A EA 201100606A EA 022480 B1 EA022480 B1 EA 022480B1
Authority
EA
Eurasian Patent Office
Prior art keywords
patterns
block
ornaments
identification
module
Prior art date
Application number
EA201100606A
Other languages
Russian (ru)
Other versions
EA201100606A1 (en
Inventor
Тельман Аббас Оглы Алиев
Гюльчин Гюльгусейн Кызы Абдуллаева
Айдын Кязим Оглы Кязим-Заде
Назакет Гаджи Кызы Курбанова
Original Assignee
Тельман Аббас Оглы Алиев
Гюльчин Гюльгусейн Кызы Абдуллаева
Айдын Кязим Оглы Кязим-Заде
Назакет Гаджи Кызы Курбанова
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 Тельман Аббас Оглы Алиев, Гюльчин Гюльгусейн Кызы Абдуллаева, Айдын Кязим Оглы Кязим-Заде, Назакет Гаджи Кызы Курбанова filed Critical Тельман Аббас Оглы Алиев
Priority to EA201100606A priority Critical patent/EA022480B1/en
Publication of EA201100606A1 publication Critical patent/EA201100606A1/en
Publication of EA022480B1 publication Critical patent/EA022480B1/en

Links

Abstract

The invention relates to information technologies and concerns systems of detection and identification of patterns and ornaments in arts and crafts articles, and in particular hand-made carpet articles. The invention essence is determination of informative parameters for comparison with reference parameters, in form of colour pixels distribution density diagrams, determination of the presence and type of symmetry, and calculation of the centre of gravity of the article. Graphical images of the patterns and ornaments are decomposed to elementary components along the contours formed at borders of pixels of different colours. In this process, depending on the colour pixels in the pattern or ornament, the colour image is changed to a black-and-white one so that if margins are black, then the contours are white, and vice versa. Decomposition in accordance with the developed algorithms is performed successively: from a complex image to simple shapes, then in colour components and main patterns. The proposed method of detection and identification of patterns and ornaments in arts and crafts articles is an intelligent-information expert system, and allows not only keeping a register of valuable arts and crafts articles, in particular, hand-made carpets, but also carrying out their identification and detection and restoration of the articles on the basis of preserved fragments, and also formation of new patterns, ornaments and carpets.

Description

The invention relates to information technology and relates to systems for recognition and identification of patterns and ornaments of arts and crafts and, in particular, handmade carpets. The invention consists in the determination of informative parameters for comparison with the reference, in the form of graphs of the densities of the distribution of color pixels, determining the presence and type of symmetry and calculating the center of gravity of the product. The decomposition of graphic images of patterns and ornaments to elementary components is carried out along contours formed on the borders of pixels of various colors, and depending on the color pixels in the pattern or ornament, the color image becomes black and white so that if the fields are black, the contours are white and vice versa. The decomposition in accordance with the developed algorithms is carried out sequentially: from a complex image to simple shapes, and then according to the color components and the basic patterns. The inventive method of recognition and identification of patterns and ornaments of artistic and applied products is an intellectual information expert system and allows not only to have a register of valuable artistic and applied products, in particular, hand-made carpets, but also to identify and recognize them and restore products from the fragments that remain, and also to form new patterns, ornaments and carpets.

022480 B1

The invention relates to information technology and relates to systems for recognition and identification of patterns and ornaments of arts and crafts and, in particular, handmade carpets.

Automation of recognition and identification of graphic images in all areas of human activity and, in particular, in the arts and crafts is one of the urgent problems of modern information technologies.

The known method (1) pattern recognition, which is a complex software package. It includes image processing of at least a part of the image (segmentation), identification parameters (attributes) identification, parameters representation (normalization) in the form of modified image data and recognition, including estimation of the scattering function based on the detection of the start and end points and comparison of the selected signs with reference estimates of the scattering point function. However, the method is intended only to recognize a black and white bar code.

The closest in technical essence is the well-known method (2) of image recognition of objects, which is a software product, including the creation of a reference database, scanning and discretization (decomposition) of an image in order to highlight an object in an image, the definition of informative parameters of an object. The selected image is represented as an array of pixels - the density of the distribution of pixels, for comparison with the reference arrays. The dimension of the array depends on the image discretization. The evaluation of the scattering function when creating an array is focused on recognizing only grayscale gradations. The method allows to accurately recognize the image and provides invariance to affine transformations of rotation and scaling. The disadvantages of this invention include the fact that the product is not intended for recognition of images in the color gamut, which is one of the most important characteristics parameters for patterns and ornaments in applied arts.

The objective of the invention is to create a way to automate the process of recognition and identification of color patterns and ornaments of arts and crafts, in particular handmade carpets.

The essence of the invention consists in the method of recognition and identification of patterns and ornaments. The method includes scanning, scaling and decomposition of the image, determination of informative parameters of the selected image and presentation of these parameters as graphs of the pixel distribution density of colors and, additionally, to identify the pattern or ornament as a whole, determine the presence and type of symmetry and calculate the center of gravity of the product. The decomposition of graphic images of patterns and ornaments to the elementary components is carried out along contours formed on the borders of pixels of various colors; however, depending on the color pixels in the pattern or ornament, the color image becomes black and white in such a way that if the fields are black, the outlines are white and vice versa. The decomposition in accordance with the developed algorithms is carried out sequentially: from a complex image to simple shapes, and then according to the color components and the basic patterns.

The invention also consists in creating an intellectual information system that implements this method. The intellectual information system (expert) consists of three modular units: a module for the certification of artistic and applied arts; a recognition and identification module, a databank module and three independent blocks: a block for assessing the presence of the Golden Section, Fibonacci series and visual centers; a block of computer-generated patterns, ornaments, and works of applied art (restoration and restoration); block the formation of educational and campaign materials. The module for the certification of products of artistic and applied art contains a block of technical and technological characteristics; block of images and block of expertise. The recognition and identification module contains a block of recognizable patterns and ornaments, a block for selecting a pattern, decomposition, information signs, a solver and a block for recognition and identification. The module of the data bank contains a catalog of works of art and applied art and a catalog of individual patterns and ornaments; however, the output of the certification module is connected to the input of the recognition and identification modules and the data bank, and the output of the recognition and identification module is connected to the input of the data bank module and the unit for evaluating the presence of the Golden Section, Fibonacci series and visual centers, the output of which is connected to the input of the database module. The outputs of the data bank module are connected to the inputs of the computer-generated unit for the formation of patterns, ornaments and articles of applied art (restoration and restoration) and the unit for the formation of educational and campaign materials. The system is logged through the passport module, and the user is logged out of the data bank module, the computer-generated block of patterns, ornaments, and applied arts (restoration and restoration) and the block for the formation of educational and campaign materials.

Comparative analysis of the proposed solution and the prototype showed that the proposed solution differs from the prototype by the following essential features: the object of the invention, graphics decomposition algorithms and informative parameters for comparing selected objects with reference ones, which are graphs of pixel density distributions of color contours, formed on the borders of pixels of different colors. Other distinctive features for identifying a product are the determination of the presence and type of symmetry and the calculation of the center of gravity of the product as a whole and / or selected patterns and ornaments. In addition, an essential feature is a developed intellectual information system that implements the inventive method. Therefore, the claimed invention meets the criterion of novelty.

Comparative analysis with other well-known solutions in this area showed that it was not found as objects of automatic recognition and identification of objects of applied art or other objects of recognition containing a large color gamut, as well as corresponding intellectual information systems that implement a similar method of recognition and identification. In the claimed invention, the decomposition method — an algorithm for decomposing a complex pattern or ornament — is performed by color pixels. At the same time, the color image becomes black and white, depending on the pixel color in the pattern or ornament. This allows you to decompose any fragment in the form of fairly simple functionals. On a plane in Cartesian coordinates, for a simple figure of the same color (Fig. 1a) this is an open loop, and for two-color (Fig. 2a) it is a closed loop (Fig. 2b) or a field (Fig. 2c). Then the pattern or ornament pattern can be represented by a complex function of contours and color, and the whole picture of the image can be represented by a graph of pixel density distribution in the form p = X (*,>y>); Λ, U /); Λ; В Р} where екк ( х 1, У]) are the outlines of separate images; ίρζΧί, γ,) - the planes of images of the same colors without gaps - internal fields; And to - the number of identical contours; In p - the number of identical planes; ρ - color pixels, k = 1, 2, ... η, ρ = 1, 2, ... t.

When And to = 1; At p = 0, an image of a simple contour (Fig. 1a) and graphs of density distributions of pixels along the axes of the CX (Fig. 1b) and RH (Fig. 1c) are obtained. When Ak = 1; BP = 1 receive the image of the pattern (Fig. 2a), its decomposition into the pattern contour (Fig. 2b) and the internal field of the pattern (Fig. 2c) and the graphs of density densities of pixels along the axes OX and OA for the contours of the pattern and its internal field (Fig 2b1, b2, b1, b2, respectively). The numerical values of A c and B p depend on the complexity of the ornament. The more complex the pattern, the more contours d and internal fields £ increases. For the complex pattern shown in FIG. 3, the size of размеромχΜ, the function of decomposition into contours and internal fields can be represented as follows:

Where

* 2, X | · - X 2> Y M / Р ( Х Ч'У ^ = * 2> Y | X 2> Y 2 · » X 2> Ui
Ή, Λ · ·· X C ') ! M, Х Р> У2 · .. x P , y ^

With this algorithm, a complex pattern can be represented as a sequence of simple shapes, which leads the task of recognizing a complex pattern to the recognition of simple shapes in black and white for storage in the reference database and matching during identification. For £ and d, respectively, plot the density distribution graphs. In addition, each pattern will be represented by its own image, the density function of the contour and middle fields. When identifying a pattern or ornament as a whole, scaling is performed as informative parameters for comparison with reference ones, the type of symmetry (3) is determined by a known method and the center of gravity of the object is calculated by a known (4) method. Thus, the developed intellectual information recognition and identification system and new features of the claimed invention create a new set of features that allows you to solve the problem, which meets the criterion of the technical level and, therefore, the claimed solution can be recognized as an invention.

The implementation of the proposed method is illustrated in FIG. 1-7. FIG. 1a shows a simple pattern of Duck, b, c - graphs of the density of the distribution of pixels of this pattern in the form of a single-color open loop along OX and OU, respectively. FIG. 2 - Leaflet pattern, a - field and contour of the pattern, b, as a result of decomposition into the contour of the pattern and field of the pattern, respectively; b1, b2, b1, b2 - graphs of the density of the distribution of the pixels of the pattern along the OH and OU for the outline of the pattern and its field, respectively FIG. 3 shows the complex Kerdekbashi pattern, on which color pixels (a, b, c, d, e) are highlighted and the density of these colors (a1, b1, b1, d1, d1) are plotted along the OX axis, and on (a2, b2 , b2, d2, d2) along the axis of the shelter. FIG. 4 is an example of a phased decomposition of a complex pattern. FIG. 5a shows the image of the carpet Mehrabi, (5b-5d) - the main color components of the carpet. FIG. 6a - the image of the Garabaglar carpet; 6b - border strips of the carpet; 6c, d, e - large patterns of the middle field of the carpet. FIG. 7 shows the scheme of functioning of the intellectual information system for recognition and identification of patterns and ornaments. The scheme contains: I - the module of certification of products of arts and crafts; II - recognition and identification module; III - databank module; 1 - a block of technical and technological characteristics; 2 - a block of images, 3 - a block of expert data; 4, 5 - blocks of recognizable patterns and ornaments; 6 - block selection pattern; 7 - decomposition block; 8 - a block of informative features; 9 - solver block; 10 - block recognition and identification; 11 - the block of the catalog of products of arts and crafts; 12 - block catalog of individual patterns and ornaments; 13 is a block for assessing the presence of the Golden Section, Fibonacci series and visual centers; 14 - computer-generated unit for the formation of patterns, ornaments and articles of applied art (restoration and restoration); 15 - block the formation of educational and campaign materials.

The method is as follows.

At the entrance of the intellectual information system to the certification module (I), a product of applied art is presented. In block 1 enter the technical and technological characteristics of the product - the geometric parameters, material and method used for the manufacture of the product. In block 2 enter the scanned image of the product. In block 3, expert data on the place of origin of the scanned product and its characteristic features are systematized. From module (I), the scanned image of the product enters the recognition and identification module (II), where it is virtually divided into structural fragments (middle field, borders, ornaments, and, if necessary, basic geometric patterns). In block 4, median patterns are collected, and in block 5 - border strips. In block 6, one of the product patterns is selected sequentially to decompose its pattern by color. In block 7, the pattern is decomposed into simple monochrome components. In block 8, informatively significant features of the pattern are determined by the presence and appearance of symmetry, the center of gravity of the figure and the distribution density of monochrome components. In block 9, the solver, using the informative parameters of the patterns obtained in block 8, classify the patterns into classes formed by experts for each type, or group of types, products. In block 10, appropriate scaling, rotation, parallel transfer are made for recognition, and the well-known (5) methods of artificial intelligence, and mathematical statistics, mathematical theory of choice and decision-making are used for identification. In block 11, according to the well-known methodology (6), the conformity of patterns and ornaments to the rules of the Golden Section, visual centers and the associated Fibonacci series are determined. The module of the data bank (III) receives all the information about the product obtained as a result of its research by the inventive system. Block 12 is a catalog of works of applied art of a certain type, and block 13 is a catalog of individual patterns and ornaments, including informative-relevant parameters for examination. Block 14 is intended for computer-generated patterns, ornaments, borders, restoration of the image of a pattern from a preserved fragment and for calculating the consumption of threads by color pixels. Block 15 is intended for the formation of multimedia materials containing information about the history of products of artistic and applied art.

An example of a specific implementation method based on the expertise of handmade carpets of Azerbaijan.

The Mehrabah carpet (Fig. 5a) or the Garabaglar carpet (Fig. 6a) is presented to the input of the intellectual information system in the certification module (I). In block 1, the technical and technological characteristics of these carpets are introduced - geometric shape, size, number of loops per square decimeter, density of knots, pile height. In block 2 enter the scanned image of the carpet. In block 3, expert data of the scanned carpet are systematized, including information about the carpet school, subdivisions of patterns and ornaments by schools, production time, basis, type of yarns, knitting patterns, patterns distribution across schools. From module (I), the scanned image of the carpet (Fig. 6a) enters the recognition and identification module (II), where it is virtually divided into structural fragments (Fig. 6b-6d) - the middle field, middle patterns, borders. In block 4, the median patterns and the middle field are assembled (Fig. 6b-e), and in block 5, the border stripes (Fig. 6b). In block 6, one of the product patterns is selected sequentially to decompose its pattern by color. In block 7 performs a decomposition of the pattern in accordance with the claimed method to simple monochrome components. The decomposition of patterns by color is carried out in stages. First, separate the external pattern on the border of different colors. Then go to the internal pattern, etc., until the pattern is broken into all components (Fig. 4a-d). The decomposition process continues until the final one-piece contour is achieved (Fig. 1a or Fig. 2a-c). In block 8, informatively significant features are determined: the presence and type of symmetry - they distinguish between mirror, rotational and helical symmetries, calculate the center of gravity of the figure and the graph of the density distribution of color pixels along the axes of ordinates. In block 9, the solver, using the informative parameters of the patterns obtained in block 8, classify them according to the subject and shape of the picture (for this example, only 32 classes (7)). In block 10, the appropriate scaling, rotation, parallel transfer are carried out for the recognition process, and the well-known (5) methods of artificial intelligence, and mathematical statistics, the mathematical theory of choice and decision-making are used for identification. In block 11, according to the well-known methodology (6), the conformity of patterns and ornaments to the rules of the Golden Section, visual centers and the associated Fibonacci series are determined. The module of the data bank (III) receives all the information about the product obtained as a result of its research by the inventive system. Block 12 is a catalog of Azerbaijani handmade carpets, and block 13 is a catalog of carpet patterns and ornaments, including informative and relevant parameters for examination. Block 14 is designed for computer-generated patterns, ornaments, borders, restoration of the image of a pattern from a preserved fragment and calculation of the thread consumption by color pixels. Block 15 is intended for the formation of multimedia materials containing information about the history of Azerbaijani carpets.

The inventive method of recognition and identification of patterns and ornaments of art-applied products is an intellectual and informational expert system and allows not only to have a register of valuable artistic and applied products, in particular handmade carpets, but also to identify and recognize them and restore the products from the fragments remaining, to form new patterns, ornaments and carpets.

Literature.

1. Application for PRF No. 2007143131 Image processing for pattern recognition, IPC S06K9 / 00, 24.05.2005.

2. PRF № 2361273, IPC O06K9 / 62 Method and device for recognition of images of objects, 12.03.07 (prototype).

3. Kochegarov BE, Industrial Design. - Vladivostok: FESTU, 2006, 153 p.

4. Fichtengolts G.M. The course of differential and integral calculus, v. 3. - M .: Fizmatlit, 2005, 728 p.

5. Hunt E. Artificial Intelligence. - M .: Mir, 1978, 558 p.

6. Stakhov A., Sluchenkova A., Shcherbakov I. The Da Vinci Code and Fibonacci Series. - Publishing house: Peter, 2006, 320 p.

7. Kerimov Latif. Azerbaijani carpet. - Baku-Len-grad: Ed. AN Az. SSR, 1961, 210 p.

Claims (5)

  1. CLAIM
    1. A method for recognizing and identifying patterns and ornaments, including scanning an image, scaling and decomposing it, determining informative parameters of a selected image for comparison with reference ones and presenting these parameters as a density of pixel distribution, characterized in that informative parameters of a selected image are density distribution graphs pixels of color, and to identify the pattern or ornament as a whole, additionally use the presence and type of symmetry and the center have a pattern or ornament; the decomposition of graphic images of patterns and ornaments is carried out sequentially from complex to elementary components along contours formed on the borders of pixels of various colors; at the same time, regardless of the color pixels in the pattern or ornament, the color image becomes black and white in such a way that if the margins are black, the outlines are white and vice versa.
  2. 2. The intellectual information system for implementing the method according to claim 1 consists of three modular units: a module for certification of artistic and applied arts, a recognition and identification module, a database module and three independent units: a block for assessing the presence of the Golden Section, Fibonacci series and visual centers, a block of computer-generated patterns, ornaments and works of applied art (restoration and restoration), a block of the formation of educational and campaign materials, with the entrance to the system have carried through the certification module, and the output of user - from the database module, the computer unit of forming patterns, ornaments and articles of applied art (restoration and recovery), and block formation of educational and promotional materials.
  3. 3. Intellectual information system according to claim 2, in which the module certification of products of arts and crafts contains a block of technical and technological characteristics, a block of images and a block of expert knowledge, with the output of the module of certification associated with the input of recognition and identification modules and the database.
  4. 4. Intellectual information system according to claim 2, in which the recognition and identification module contains a block of recognizable patterns and ornaments, blocks of pattern selection, decomposition, informational signs, solver and recognition and identification block, while the output of the recognition and identification module is connected to the input the database module and the unit for assessing the presence of the Golden Section, Fibonacci series and visual centers, the output of which is connected to the input of the database module.
  5. 5. Intellectual information system according to claim 2, in which the database module contains a catalog of products of artistic and applied art and a catalog of individual patterns and ornaments, and the outputs of the database module are connected to the inputs of a block of computer-generated patterns, ornaments and articles of applied art (restoration and restoration) and block formation
    - 4 022480 training and campaign materials.
EA201100606A 2010-12-23 2010-12-23 The method of recognition and identification of patterns and ornaments and intellectual information system for its implementation EA022480B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EA201100606A EA022480B1 (en) 2010-12-23 2010-12-23 The method of recognition and identification of patterns and ornaments and intellectual information system for its implementation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
EA201100606A EA022480B1 (en) 2010-12-23 2010-12-23 The method of recognition and identification of patterns and ornaments and intellectual information system for its implementation

Publications (2)

Publication Number Publication Date
EA201100606A1 EA201100606A1 (en) 2012-06-29
EA022480B1 true EA022480B1 (en) 2016-01-29

Family

ID=46614781

Family Applications (1)

Application Number Title Priority Date Filing Date
EA201100606A EA022480B1 (en) 2010-12-23 2010-12-23 The method of recognition and identification of patterns and ornaments and intellectual information system for its implementation

Country Status (1)

Country Link
EA (1) EA022480B1 (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5546475A (en) * 1994-04-29 1996-08-13 International Business Machines Corporation Produce recognition system
RU2163394C2 (en) * 1999-03-29 2001-02-20 Федеральный научно-производственный центр "Научно-исследовательский институт комплексных испытаний оптико-электронных приборов и систем ВНЦ "ГОИ им. С.И. Вавилова" Material entity identification method
RU63086U1 (en) * 2007-01-11 2007-05-10 Общество с ограниченной ответственностью "Ви Ай Пи Колор" Determining device characteristics tsvetostrukturnyh frame for a given image area
RU2361273C2 (en) * 2007-03-12 2009-07-10 Государственное образовательное учреждение высшего профессионального образования Курский государственный технический университет Method and device for identifying object images
EP2154631A2 (en) * 2008-08-14 2010-02-17 Xerox Corporation System and method for object class localization and semantic class based image segmentation

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5546475A (en) * 1994-04-29 1996-08-13 International Business Machines Corporation Produce recognition system
RU2163394C2 (en) * 1999-03-29 2001-02-20 Федеральный научно-производственный центр "Научно-исследовательский институт комплексных испытаний оптико-электронных приборов и систем ВНЦ "ГОИ им. С.И. Вавилова" Material entity identification method
RU63086U1 (en) * 2007-01-11 2007-05-10 Общество с ограниченной ответственностью "Ви Ай Пи Колор" Determining device characteristics tsvetostrukturnyh frame for a given image area
RU2361273C2 (en) * 2007-03-12 2009-07-10 Государственное образовательное учреждение высшего профессионального образования Курский государственный технический университет Method and device for identifying object images
EP2154631A2 (en) * 2008-08-14 2010-02-17 Xerox Corporation System and method for object class localization and semantic class based image segmentation

Also Published As

Publication number Publication date
EA201100606A1 (en) 2012-06-29

Similar Documents

Publication Publication Date Title
Cireşan et al. Multi-column deep neural networks for image classification
Hassanien et al. Rough sets and near sets in medical imaging: a review
Ullman et al. A fragment-based approach to object representation and classification
Sharan et al. Recognizing materials using perceptually inspired features
KR100819960B1 (en) Analysis of geometric surfaces by conformal structure
Huang et al. An SVM ensemble approach combining spectral, structural, and semantic features for the classification of high-resolution remotely sensed imagery
Logeswari et al. An improved implementation of brain tumor detection using segmentation based on soft computing
Eslami et al. Attend, infer, repeat: Fast scene understanding with generative models
Clausi K-means Iterative Fisher (KIF) unsupervised clustering algorithm applied to image texture segmentation
Eslami et al. The shape boltzmann machine: a strong model of object shape
Zhang et al. Tensor discriminative locality alignment for hyperspectral image spectral–spatial feature extraction
Wooding Eye movements of large populations: II. Deriving regions of interest, coverage, and similarity using fixation maps
Sandler et al. Nonnegative matrix factorization with earth mover's distance metric for image analysis
Kubovy et al. The whole is equal to the sum of its parts: A probabilistic model of grouping by proximity and similarity in regular patterns.
Hyvärinen et al. Natural image statistics: A probabilistic approach to early computational vision.
JP2002543525A (en) The method for nonlinear mapping multidimensional data, systems, and computer program products
JP2005234810A (en) Image processor
Haralick et al. A spatial clustering procedure for multi-image data
Tsogkas et al. Learning-based symmetry detection in natural images
WO2014133756A2 (en) Method and apparatus for learning-enhanced altas-based auto-segmentation
Di Zenzo Advances in image segmentation
Ullman et al. Object classification using a fragment-based representation
Li et al. The wrinkle generation method for facial reconstruction based on extraction of partition wrinkle line features and fractal interpolation
Silva et al. Classification of segments in PolSAR imagery by minimum stochastic distances between Wishart distributions
Ritter et al. Autonomous single-pass endmember approximation using lattice auto-associative memories

Legal Events

Date Code Title Description
MM4A Lapse of a eurasian patent due to non-payment of renewal fees within the time limit in the following designated state(s)

Designated state(s): AM AZ BY KZ KG MD TJ TM RU