Method and algorithm of the

PROCEEDINGS OF SPIE

SPIEDigitalLibrary.org/conference-proceedings-of-spie

Method and algorithm of the

piecewise-hyperplane clusterization using tools of pseudo-inverse

matrices

Iurii V. Krak, Hrygorii I. Kudin, Olexander V. Barmak, Eduard O. Manziuk, Andrzej Smolarz, et al.

Iurii V. Krak, Hrygorii I. Kudin, Olexander V. Barmak, Eduard O. Manziuk, Andrzej Smolarz, Orken Mamyrbaev, "Method and algorithm of the piecewise- hyperplane clusterization using tools of pseudo-inverse matrices," Proc. SPIE 11176, Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2019, 1117628 (6 November 2019); doi:

10.1117/12.2537417

Event: Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2019, 2019, Wilga, Poland

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

Method and algorithm of the piecewise-hyperplane clusterization using tools of pseudo-inverse matrices

Iurii V. Krak*

^ab

, Hrygorii I. Kudin

^b

, Olexander V. Barmak

^c

, Eduard O. Manziuk

^c

, Andrzej Smolarz

^d

, Orken Mamyrbaev

^e

a

Taras Shevchenko National University of Kyiv, Ukraine;

^b

Glushkov Institute of Cybernetics of NAS of Ukraine;

^c

National University of Khmelnytsky, Ukraine;

^d

Lublin University of Technology,

Poland;

^e

Institute Information and Computational Technologies CS MES RK, Kazakhstan

ABSTRACT

Method and algorithm of multidimensional information scaling of the characteristics features based on the results of the theory of perturbation of pseudo-inverse and projective matrices and solutions of systems of linear algebraic equations is proposed. The algorithm of a piecewise-hyperplane clusterization with the verification of a given criterion for the effectiveness of the proposed method of the clusterization is developed. An example of using the method of scaling characteristic features for recognizing the fingerspelling alphabet of the sign language is given.

Keywords: classification, clusterization, pseudo-inverse matrixes, sign language

1. INTRODUCTION

One of the important problems of classification and clusterization of information is the problem of minimizing the dimension of the space of features, the choice of criteria for optimal solutions in the processes of practical use. Such problems are effectively solved by the method of multidimensional scaling of empirical data on the proximity of objects, with the help of which the dimension of the space of essential characteristics of the measured objects is determined and the configuration of points - objects in this space is constructed. This space is a multidimensional scale, similar to the commonly used scales in various applications in the sense that the values of the specially generated essential characteristics of the measured objects correspond to certain positions on the axes of the new space ^1,10.

The purpose of this work is the development of mathematical methods for the synthesis of systems for solving problems of classification and clusterization based on information about the characteristic features of objects. These problems are proposed to be solved by constructing hyperplanes in a space derived from the original space of features using theory of perturbation of pseudo-inverse and projection matrices and solving systems of linear algebraic equations ^{2, 3}. The paper proposes a method for synthesizing a piecewise hyperplane cluster to isolate the most effective characteristic features and an algorithm for constructing piecewise hyperplane clusters that allow you to find an effective solution to the problems. The performance and efficiency of the proposed approach is shown on the example of the scaling of characteristic features for recognizing the letters of the fingerprint alphabet of the sign language ^11,12.

2. METHOD OF THE PIECEWISE-HYPERPLANE CLUSTER S Y NTHESIS

The problem of synthesis of a piecewise hyperplane cluster for a training sample of vectors

 

0 x x j: ( ) E^m,j 1,n

    , where x(1), , ( )x n - vectors from the Euclidean space of features E^m is to build a cluster so that the training sample points in this space are located quite close, in the sense of a given distance criterion, to some set of hyperplanes that are formed according to this sample ^13,14.

Note that in the formulation of this problem for clusterization of information, the components of the set of hyperplanes are not known in advance. Therefore, for the correct construction of piecewise hyperplane clusterization procedures, it is assumed that vectors x(1),,x(n)from the space of features E^m can belong to one of several hyperplanes

* krak@univ.kiev.ua

Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2019, edited by Ryszard S. Romaniuk, Maciej Linczuk, Proc. of SPIE Vol. 11176, 1117628

© 2019 SPIE · CCC code: 0277-786X/19/$21 · doi: 10.1117/12.2537417

Proc. of SPIE Vol. 11176 1117628-1 Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

)), ( ), ( (Ak bk

L where A(k)E^sm, b(k)E^s, k1,2,... some given dimension s, (sm). Here A(k) and b(k) are matrix and vector parameters, respectively, for a fixed hyperplaneL(A(k),b(k)) k1,2,... .

The proposed method of synthesizing a piecewise-hyperplane cluster is based on using the representation of hyperplanes by means of a set of solutions (pseudo solutions) of systems of algebraic equations.

( ) ( )

A k xb k , (1)

( ( ), ( )) { ⁿ: ( ) ( ) ( ( )) , ⁿ}

L A k b k  xE xA k b k^ Z A k z zE . (2) Here A^ denotes a pseudo inverse matrix, Z is a projection matrix.

Let us give some mathematical results on the inversion (pseudo inversion) of matrices and the construction of projection matrices, which are important for solving the problem of synthesizing a piecewise-hyperplane cluster. Let the matrix be given A(a_ij),i1,m j1,n and we write down, which are important in further studies, the representations of this matrix in columns or rows, respectively:

( (1) ... ( )), ( ) ⁿ, 1, , A a a n a j E j n



(1)^T ( )^T_m



^T, ( )^T_i ⁿ, 1, A a a a E i m, where Tdenotes the transpose symbol.

We consider the singular decomposition of an arbitrary matrix A of dimension mn of rank rmin(m,n) in the form

1 r

T i i i i

A u v







^{, where: 12 ... r20} - non-zero eigenvalues of matrices AA^T,A^TA; v_iEⁿ,i1,r - orthonormal set of eigenvectors of the matrix A^TA, which correspond to non-zero eigenvalues _i²,i1,r, A Av^T _i_i²v_i,

1, , ^T_{i j ij}

i r v v  u_iE^m,i1,r - orthonormal set of eigenvectors of the matrix AA^T, which also correspond to non- zero eigenvalues _i²,i1,r,AA u^T _i_i²u_i, 1, , i ,

T

j ij

i r u u  _ij is the Kronecker’s function.

Let us give the definition of a pseudo inverse matrix in the Penrose optimization form ⁴. For a matrix AE^mn, a pseudo inverse matrix A^E^mn is defined by the relation:

2 ( )

arg min .

A m

x b

b E A b^ x

 

  

Here ( ) min ²

A n

x E

b Arg Ax b SS

    . Also, using the singular representation of the matrix AE^{m n} S, the pseudo inverse matrix A^E^{n m} S can be represented as ⁵: ¹

1 n

T j j j i

A^ x y ^S







^.

Let us consider matrices important for practical application, that are defined and calculated using matrices A and A^: 1) a projection matrix

1

( )

r T i i j

P A A A^ x x S



 



that is an orthogonal projector onto the subspace T

LA generated by the row vectors of the matrix AS;

2) projection matrix Z A( )I_nP A S( ) - orthogonal projection onto a subspace orthogonal to a subspace T

LA , In - unit matrix;

3) matrix













 

 ^r

j

j j jx x A

A A R

1

) 2

( )

(  .

Proc. of SPIE Vol. 11176 1117628-2 Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

Note the important properties of projection matrices P and Z : P A( )Z A( )I_n, P A( )A A^ ,

1

( )

r T i i i

P A v v







^,

1

( )

r T i i i r

Z A v v

 





^.

In addition to P- and Z - projectors (projection matrices) associated with the basic subspaces of the matrix A, matrices )

(A

R are of important application and defined according to the formulas: ( )R A A A^( ^)^T, ²

1

( )

r T i i i i

R A ^v v







^.

Note that the calculation of the pseudo inverse matrix for an arbitrary matrix can be reduced to the calculation of the corresponding to it some inverse matrix using the following relations: A^(A A A^T )^{ T} A AA^T( ^T)^,A^ R A A( ) ^T. Using the above mathematical ratios for the inversion and pseudo inversion of matrices, we write the necessary formulas for the synthesis of a piecewise hyperplane cluster.

The distance 



^x(j),L(A,b)



from the point x(j) to the hyperplane L(A,b) will be found from the relation

 

²

2 x j L A b( ), ( , ) A b( Ax j( ))

  ^  , x² x x^T .

To calculate the sum of the squares of the distances of the set of points x(j), j1,n to the hyperplane L(A,b), we use the following formula:

 

  ^{      }

2

1 1

: ( ), 1, , ( , ) ( ) ( ) ( ) ( ) ( ) ( )

n T n

T T T

j j

x x j j n L A b b Ax j R A b Ax j tr R A b Ax j b Ax j



 

 



  



  ^.

Here ()tr - matrix trace.

Then, for given values A, x(j), j1,n, the optimal value of the vector of the right-hand side of the system of equations determining the hyperplane is determined from the conditions

 

 

2

1

ˆ arg min : ( ), 1, , ( , ) ,ˆ 1 ( ).

S

n

opt b R

j

b Ax x x j j n L A b x x j

 n

 

   



From here, the distance ^

 

^{x x j: ( ), j1,n}



^{,L A b( , )}



of the set of points ( ) ,x j j1,n to the hyperplane, with an optimal vector b_opt, is calculated by the following formula:

 



x : x(j), j ¹,n ^,L A b^{( , opt( ))}A



⁽trA AXX^{T) ,12}

   ^

where X ( (1)) .. ( )),x x n x j( )x j( )xˆ, j1,n.

The optimal matrix AE^smis defined as a solution of the problem

 

2 ,

arg _T min _{s m} ( : ( ), 1, ,.

s

opt AA E A E

A _  x x j j n

 

 



1



( , _opt( ))) ^T_{m s} ^T_m ^T. L A b A  u _{ } u

Wherein ²

1 r T

opt opt j

j m s

trA A XX^ 

  





^{,( ,u1 ,um) ( ,T u1 ,um)Im.}

If the hyperplanes are defined using the singular decomposition of the matrix

Proc. of SPIE Vol. 11176 1117628-3 Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

r

i 1

X u_i_{i i}v^T







^{, Us(u1 ...us),}

where s - is the number of higher u-vectors (corresponding to large singular numbers) of singular decomposition, then for this case the distance (x(i),L(A,b)) is determined by the relations:

2

1

ˆ ˆ ˆ ˆ ˆ

( ( ), ( , )) ( ( ) ) ( )( ( ) ) || ( ) || ( ( ) ) ( ( ) ).

s

T T T T

s j j

j

x i L A b x i x Z U x i x x i x x i x u u x i x





      





Using the above results on pseudo inversion of matrices and calculation of distances, a piecewise hyperplane clusterization method is proposed. The idea of the method is to perform a sequence of steps, at each of which the parameters of the hyperplanes L(A(k),b(k)), k1,2,... are found. These hyperplanes are constructed within the framework of performance the requirements of the efficiency criterion of the implemented hyperplane clusterization. As initial parameters of piecewise hyperplane clusterization, it is assumed that all the training set vectors x(1),,x(n) from the space of features E^m are optimally approximated by the hyperplane L(A_opt(1),b_opt(1)). At these values A_opt(1),b_opt(1) , the performance efficiency criterion of hyperplane clustering is checked. If the efficiency criterion is satisfied, this means that the result of the construction of the piecewise hyperplane cluster is achieved by building the cluster with one hyperplane and those characteristic features are the optimal set. If the conditions of the efficiency criterion in the framework of this (first) hyperplane are not met, then the transition to the construction of the second hyperplane of the cluster is carried out. For this purpose, the vectors that determine the non-fulfillment of the efficiency criterion are excluded from the training sample, that is, a subset is formed (1)



x:x(j1)E^m,j11,n1



. Then the actions on the optimal approximation of the subset ¹ by the hyperplane L(A_opt(2),b_opt(2)) are repeated, and at the new optimal values A_opt(2),b_opt(2), the fulfillment of the criterion of clusterization efficiency is checked. When the criterion is met, the result of the construction of the piecewise hyperplane cluster is reached, then the resulting vectors of characteristic features are added to the optimal set of features. When not performing - it is necessary to exclude from the training sample those vectors that cause to the non-fulfillment of the efficiency criterion and repeat the procedure. Note that this finite recurrent process will always ensure the construction of an optimal piecewise hyperplane cluster ^15,16.

3. ALGORITHM OF S Y NTHESIS THE PIECEWISE-HYPERPLANE CLUSTER

Using the proposed method of synthesizing piecewise hyperplane clusters, the algorithm of such a synthesis can be represented as the following sequence of actions:

1. Formation of a single-link cluster (the number of a link in a cluster is an index in brackets):

1) All vectors of the training sample ^(0)



^{x:x(j)Em,j1,n}



from the space of features are to be optimally approximated by a hyperplane L(A_opt(1),b_opt(1)), which is defined as the set of solutions (pseudo-solutions) of systems of algebraic equations (1), (2), where A(k) and b(k) are respectively the matrix and vector parameters of a certain hyperplane (basic formulas for constructing a hyperplane cluster are implemented) ^16,17.

2) To form a set (1)



x:x(j1)E^m,j11,n1



of points (0) for which the condition is satisfied, namely:

1 1 1 1 1 1 1 min

(b_opt(1)A_opt(1) ( ))x j ^TR A( ^T_opt(1)) ( b_opt(1)A_opt(1) ( ))x j h ,

where h_min - valid distance of vectors from components of the cluster of hyperplanes. The linear dependence or independence of vectors that can be removed from the set makes it possible to simplify the form of the formulas for the distances of these vectors from the corresponding hyperplanes.

3) The stop of the algorithm at the stage of the first cluster link can be completed after the distance of each of the vectors of the training sample ^(0)



^{x:x(j)Em,j1,n}



to the hyperplane L(A_opt(1),b_opt(1)) does not exceed the allowable distance ^18,19.

Proc. of SPIE Vol. 11176 1117628-4 Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

4) If the set (1)



x:x(j1)E^m,j11,n1



includes at least two vectors, then go on to form the second link of the cluster.

2. Formation of the second cluster link. Consists of the following:

1) From the set obtained in the process of building the first link(1)



x:x(j1)E^m,j11,n1



, a hyperplane ))

2 ( ), 2 ( (A_opt b_opt

L defined as the set of solutions (pseudo - solutions) of a system of algebraic equations

(2) (2),

k k

A xb k1, 2,... (3)

is constructed.

2) To calculate optimal A_kopt(2), b_kopt(2) for (L A_k(2),b_k(2)), k1,2,...

3) To form a set ⁰_k(2), k1,2,...by the distance of the vector x(j)⁰(2) to each of the hyperplanes (3):

     

2 x x j: ( ), L A( _kopt(2),b_kopt(2)) b_kopt(2) A_kopt(2) ( )x j ^TR^T(A_kopt(2)) b_kopt(2) A_kopt(2) ( )x j

    

i.e., perform actions similar to paragraph 3 of the construction of the first link.

4) To go to step 2 with new subsets ₁^j(1), ₂^j(1), j1,2,. The superscript j denotes the number of iterations at the second link stage ^20,21.

The algorithm is stopped at the stage of the second cluster link after the distances of each of the vectors of the corresponding partitioning element to the corresponding hyperplane are not improved.

The efficiency criterion of the carried out hyperplane clustering is checked (for example, the requirement of the level of compactness of the cluster links). When the criterion is fulfilled, the cluster is completed, and if it is unfulfilled, the transition to the formation of the third cluster link is carried out. Then the process repeats and, since it is finite, we will always find a solution to the clustering problem ^22,23.

4. FINGERSPELLING ALPHABETS RECOGN I TION BASED ON PIECEWISE- HYPERPLANE CLUSTERIZATION

To test the effectiveness of the proposed approach of the piecewise-hyperplane clusterization we used characteristic features for recognizing the ductile letters of the alphabet of the Ukrainian sign language ^6-9. As characteristic features, 52 features were taken, which were divided into 6 groups, depending on the method of their getting. The construction of piecewise hyperplane clusters was carried out with groups of features that characterize the geometric - topological parameters of the human hand when showing the letters of the dactyl alphabet and for which an acceptable recognition quality was obtained ⁶. Using the example of the classification of nine dactylemes (А, Б, В, Г, Ж, І, Є, И, Й) according to three and five characteristic features, the separation of these dactylemes on the scaling plane was obtained. Wherein three characteristic features were taken: compactness, directionality, elongation, and in experiments with five features:

the ratio of width to height, four angles values between vectors drawn from the center of the hand to the most distant points of the hand ^24,25.

Experimental results using these three characteristic features are given in Fig. 1 where image of the location of the dactylemes on the scaling plane, normalized to the interval of 0,1. Without reducing the generality of research, dactylem A was placed at the origin ²⁶.

The results of clusterization with the use of five characteristic features are shown in Fig. 2., where the dactylem A was also placed at the origin.

Proc. of SPIE Vol. 11176 1117628-5 Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

Fig.1 – Distances dactylemes from hyperplane of scaling for three characteristic features.

Fig.2 – Distances dactylemes from hyperplane of scaling for five characteristic features.

5. CONCLUSION

The paper proposes a method and algorithm for multidimensional scaling of the characteristic features of objects of recognition using the means of matrix pseudo inversion, building piecewise hyperplane clusters that provide a solution of the problem. The proposed method makes it possible to analyze information about the set of characteristic features and to identify those that are essential for solving the recognition problem, which is important due to their significant number and for difficult separable classes. The effectiveness of the proposed approach is shown by the example of obtaining clusters for dactylemes gesture language in order to obtain the optimal number of characteristic features for effective recognition.

REFERENCES

[1] Devison, M., [Multidimensional scaling], Finansy I Statistyka, Moscow (1988).

[2] Kirichenko, N. F., Krak, Yu. V. and Polishchuk, A. A., “Pseudo inverse and projection matrices in problems of synthesis of functional transformers,” Cybernetics and Systems Analysis 40(3), 407-419 (2004).

Proc. of SPIE Vol. 11176 1117628-6 Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

[3] Kirichenko, N. F. and Kudin, G. I., “Analysis and synthesis of signal classification systems by perturbing pseudoinverse and projection operations,” Cybernetics and Systems Analysis 45(4), 613-622 (2009).

[4] Penrose, R., “A generalized inverse for matrices, “ Proceeding of the Cambridge Philosophical Society 51., 406-413 (1955).

[5] Ben-Israel, A. and Greville, T. N. E., [Generalized inverse: Theory and Applications. 2-nd Ed.] Springer-Verlag, New York (2003).

[6] Kryvonos, Iu. G., Krak, Iu. V., Barmak, O. V. and Shkilniuk, D. V., “Construction and identification of elements of sign communication,” Cybernetics and Systems Analysis 49(2), 163–172 (2013).

[7] Krak, Iu. V., Golik, A. A. and Kasianiuk, V. S., “Recognition of dactylemes of Ukrainian sign language based on the geometric characteristics of hand contours defects,” Journal of Automation and Information Sciences 48(4), 90- 98 (2016).

[8] Kryvonos, Iu. G., Krak, Iu. V., Barmak, O. V. and Kulias, A. I., “Methods to Create Systems for the Analysis and Synthesis of Communicative Information,” Cybernetics and Systems Analysis 53(6), 847–856 (2017).

[9] Krak, Iu. V., Kryvonos, I. G., Barmak, O. V. and Ternov, A. S., “An Approach to the Determination of Efficient Features and Synthesis of an Optimal Band-Separating Classifier of Dactyl Elements of Sign Language,”

Cybernetics and Systems Analysis 52(2), 173–180 (2016).

[10] Zabolotna, N. I., Pavlov, S. V., Radchenko, K. O., Stasenko, V. A., Wójcik, W. and Kussambayeva, N., “Diagnostic efficiency of Mueller-matrix polarization reconstruction system of the phase structure of liver tissue,” Proc. SPIE 9816 (2015).

[11] Kukharchuk, V. V., Bogachuk, V. V., Hraniak, V. F., Wójcik, W., Suleimenov, B. and Karnakova, G., “Method of magneto-elastic control of mechanic rigidity in assemblies of hydropower units,” Proc. SPIE 10445 (2017).

[12] Vyatkin, S. I., Romanyuk, A. N., Gotra, Z. Y.; Romanyuk, O. V., Wójcik, W., Romaniuk, R., Amirgaliyev, Y. and Assembay, A., “Offsetting, relations and blending with perturbation functions,” Proc. SPIE 10445 (2017).

[13] Wojcik, W., Komada, P., Cieszczyk, S. and Firago, V. A., “ECTL application for carbon monoxide measurements,”

Proc. SPIE 5958 (2005).

[14] Wojcik, W., Lach, Z., Smolarz, A. and Zyska, T., “Power supply via optical fibre in home telematic networks,”

Przegląd Elektrotechniczny 84(3), 277-279 (2008).

[15] Wojcik, W. and Kisala, P., “The application of inverse analysis in strain distribution recovery using the fibre bragg grating sensors,” Metrology and Measurement Systems 16(4), 649-660 (2009).

[16] Wojcik, W., Gromaszek, K., Kotyra, A., and Ławicki, T., “Pulverized coal combustion boiler efficient control,”

Przegląd Elektrotechniczny 88(11B) 316-319 (2012).

[17] Zabolotna, N. I., Wojcik, W., Pavlov, S. V. Ushenko, O. G., and Suleimenov B., “Diagnostics of pathologically changed birefringent networks by means of phase Mueller matrix tomography,” Proc. SPIE 8698, 86980C (2013).

[18] Chepurna, O., Shton, I., Kholin, V., Voytsehovich, V., Popov, V., Pavlov, S., Gamaleia, N., Wójcik, W. and Zhassandykyzy, M. “ Photodynamic therapy with laser scanning mode of tumor irradiation,” Proc. SPIE 9816, 98161F (2015).

[19] Krak, Y. V., Barmak, A. V., Bagriy, R. A. and Stelya, I. O., “Text entry system for alternative speech communications,” Journal of Automation and Information Sciences 49(1), 65–75 (2017).

[20] Kryvonos, Iu. G., Krak, Iu. V., Barmak, O. V. and Bagriy, R. O., “Predictive text typing system for the Ukrainian language,” Cybernetics and systems analysis 53(4), 495–502 (2017).

[21] Zabolotna, N. I., Pavlov, S. V., Ushenko, A. G., Sobko, O. V. and Savich, V. O., “Multivariate system of polarization tomography of biological crystals birefringence networks,” Proc. SPIE 9166 (2014).

[22] Krupa, A. and Sawicki, B., "Massive simulations using mapreduce model," Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Srodowiska – IAPGOS 4, 45-47, (2015).

[23] Vassilenko, V., Valtchev, S., Teixeira, J. P. and Pavlov, S., “Energy harvesting: an interesting topic for education programs in engineering specialities,” Proc. Internet, Education, Science, 149-156 (2016).

[24] Kvyetnyy, R., Bunyak, Y., Sofina, O., Kotyra, A., Romaniuk, R. S. and Tuleshova, A., "Blur recognition using second fundamental form of image surface", Proc. SPIE 9816 (2015).

[25] Kvyetnyy, R. N., Romanyuk, O. N., Titarchuk, E. O. Gromaszek, K. and Mussabekov, N., "Usage of the hybrid encryption in a cloud instant messages exchange system," Proc. SPIE 10031 (2016).

[26] Kvyetnyy R., Sofina, O., Orlyk, P., Utreras A. J., Smolarz, A., Wójcik, W. and Orazalieva, S., "Improving the quality perception of digital images using modified method of the eye aberration correction," Proc. SPIE 10031, 1003113 (2016).

Proc. of SPIE Vol. 11176 1117628-7 Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 07 Nov 2019

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use