• Nu S-Au Găsit Rezultate

View of Face Local and Global Features Recognition based on Using Hybrid Graph based Cosine Transform

N/A
N/A
Protected

Academic year: 2022

Share "View of Face Local and Global Features Recognition based on Using Hybrid Graph based Cosine Transform"

Copied!
14
0
0

Text complet

(1)

Face Local and Global Features Recognition based on Using Hybrid Graph based Cosine Transform

Leta Tesfaye Jule1, A. Sampath Kumar2, Krishnaraj Ramaswamy3

1Centre for Excellence in Indigenous Knowledge, Innovative Technology Transfer and Enterpreneurship and, Department of Physics, College of Natural and Computational Science,

Dambi Dollo University, Ethiopia.

2 Department of Computer Science and Engineering, Dambi Dollo University, Ethiopia..

3Centre for Excellence in Indigenous Knowledge, Innovative Technology Transfer and Enterpreneurship and, Department of Mechanical Engineering, College of Natural and

Computational Science, Dambi Dollo University, Ethiopia.

ABSTRACT

Face Recognition utilizing Hybrid graph based Cosine Transform(HG-DCT) for Global and Local Features, from the information base it perceives the relating face picture. Our point is to extricate neighborhood highlights as of a given front facing face. Nearby highlights are right eye, left eye, mouth and nose. The above neighborhood highlights are extricated physically.

Mixture chart based Cosine Transform (HG-DCT) is practiced by every one of the nearby highlights separately and furthermore to the worldwide highlights. At long last, the outcomes acquired in the two cases will be thought about. Neighborhood highlights like nose, mouth and eyes are separated physically obtained from the provided picture of face. Picture is utilized as an info. Neighborhood include extraction, particularly eye extraction, diminishes the impact of changing qualities like posture, articulations and so on the face acknowledgment framework.

Both nearby and worldwide highlights are utilized for examination. By contrasting the positions for worldwide and neighborhood includes, the bogus acknowledgment rate for HG-DCT can be limited. For all the images from database the standardization and zig-zag scanning approach is maintained to reduce the vectorization component and hence correlation is achieved by utilizing the Euclidian distance method. The proposed HG-DCT is compared with two existing methods such as ANFIS and ANF method in terms of accuracy, precision, recall, and F1 – Score. It is find that the proposed HG-DCT achieves 90%of accuracy, 89% precision, 88% recall and 87% of f1- score.

1. Introduction

A face confirmation framework is on a very basic level an application expected to see or check an individual either from an automated picture or in the format of video got as a source of video [1]. However, another biometric dependable system for particular existing ID, like momentous engraving assessment or channeling of iris, the above techniques ordinarily depend upon the collaboration of the people, while an individual prominent affirmation structure subject to appraisal of forward looking or face’s profile photographs is routinely reasonable shorn of the

(2)

part's speculation or mediation [2]. The different methods to alter perceiving confirmation or check was done through separating picked facial highlights from a facial instructive file and the picture. The strategies are generally utilized in frameworks of security [3].

For example, this progression should be utilized like ATM's prosperity effort; rather than utilizing a debit or credit card or individual ID, face picture of the individual, separate it by their photograph in the bank enlightening list to affirm the character of the appropriate individual was obtained by the ATM [4]. On relative outlines, the above idea might in like way be inferred to PCs; through utilizing a web camera to get an individual’s general picture, generally utilized secret word was substituted by the face as a way to deal with sign in and hence, support oneself.

Given a goliath information base of pictures and a photo, the issue is to peruse the educational list a little strategy of records with a definitive target that one of the picture records worked with the photo [5]. The achievement of the strategy could be evaluated comparably as the degree of the fitting response once-over for the records measured in the instructive list [6].

Insistence issue was badly arranged through incredible fluctuation in pivot, head and slope, force of lighting and point, look, creating, and so forth Some changed endeavors at facial certification by machines have pondered fundamentally no alterability in these aggregates. An overall assertion of the machine confirmation issue of countenances was outlined by: provided a scene’s video picture or still, perceive or certify at any rate one people in the scene utilizing a put away instructive record of appearances. Open assurance data like race, age, sexual course, look, or talk might be utilized in narrowing the pursuit [7]. The reaction for the issue consolidates division of faces, include extraction from face locales, certification, or confirmation.

In unmistakable check issues, the obligation to the framework is a dull face, and the construction reports back the picked character from an instructive list of known people, while in attestation issues, the design needs to demand or pardon the pronounced character of the information face [8].

A part of the face affirmation’s numerous usages consolidates development, public ID, driving licenses, recognizable proof, balloter enrollment, security application, clinical records, singular device login, work territory login, human-robot-affiliation, human-PC association, astute cards, etc. One of its applications is Criminal Face Identification framework. Setting up the character of any criminal by building the picture of the criminal dependent on data given by onlookers and contrasting it and existing criminal pictures in the information base accessible with exploring offices [9]. Development of the picture of the criminal utilizing divided/cut pictures of human face like mouth, eyes, nose, etc.is accessible in the data set with the assistance of onlookers [10].

(3)

1.1 THERE ARE THREE APPROACHES FOR FACE RECOGNITON Comprehensive coordinating with techniques:

These techniques utilize the entire face locale as a crude contribution to the acknowledgment framework. Perhaps the most broadly utilized portrayals of the face locale is Eigen pictures, which is innately founded on head part examination.

Highlight based coordinating with techniques:

For the most part, in the above techniques, highlights of neighborhood such as nose, mouth and eyes are first extracted and its regions and close perceptions are taken care of as contributions to a classifier.

Mixture techniques:

It utilizes both neighborhood highlights and entire face area to perceive a face. This technique might actually propose the best among the 2 sorts of strategies [11].

Our point separates nearby highlights from the provided by the face which is faced front. Face picture got from the client is edited to such an extent that lone the front facing face picture is extricated, dispensing with the foundation. The picture is limited 128 × 128 size pixels. Each and every pictures in the data set were dark level pictures. HG-DCT was practiced by the whole picture. This gives coefficients of HG-DCT, that were the worldwide highlights.

2. Literature Survey

In [12], Neuro-fuzzy (NF) combination in a multimodal face acknowledgment utilizing ICA, SIFT and PCA is presented. In this work, multimodal face acknowledgment is talked about and the carried out of with NF blend. The chief segment investigation (PCA) and free segment examination (ICA) just as highlight extraction dependent on SIFT are utilized. The acknowledgment ID decide dependent on NF surmising framework. In [13], face acknowledgment utilizing neuro-fluffy and eigenface is presented. In this work, a human presence is recognized by removing the skin territory by utilizing the Eigen worth of face technique. At that point bu utilizing a neuro-fluffy technique, the face is perceived. In [14], face acknowledgment framework utilizing versatile neuro fluffy deduction framework (ANFIS) is presented. In this work, ANFIS with PCA calculation has been proposed by considering various commitments of the preparation tests. In [15] face acknowledgment (FR) in light of decision level combination is presented. In this paper, another technique named C2D CNN is proposed. In [16] facial acknowledgment dependent on versatile neuro fluffy (ANF) derivation framework is presented. In this work, the principle commitment depends on include extraction and characterization. A tale approach was received for breaking down the looks with the revelation of some regular data from a few articulations. The general articulations and the particular

(4)

articulations are recognized through the normal and specific patches [17]. A methodology relying on the staggered Haar wavelet was utilized to extricate the highlights of the appearance from the conspicuous face areas on different scales [18]. Another element extraction foundation which helped in meaning the distinctions in the look as a direct combination of restricted premise capacities was introduced in [19].The creators reported a novel methodology for separating the recognizable spaces of the face with Self-Organizing Map based Neural Network classifier. An epic methodology was introduced to perceive the look with three stages, for example, highlight extraction, include advancement and feeling acknowledgment [20]. The face portrayals are initially separated utilizing modified Local Binary Patterns. The looks are isolated utilizing implanted PSO calculation through tracking down the significant and discriminative highlights.

Various classifiers are utilized for perceiving the feelings like cheerful, dismal, outrage, shock, dread, disdain and unbiased. This methodology is evaluated utilizing broadened Cohn Kanade and MMI information bases. The upside of this work is computational expense and speed of the intermingling. The creator acquainted a novel methodology with find the explicitness of the appearance variety in the face [21]. The particularity of the articulation was connoted through the trio insightful articulation acknowledgment as per Action Unit (AU) and fix weight advancement. To procure better speculation, the dynamic AUs are identified and introduced in testing tests utilizing meager portrayal based methodology. The exactness is upgraded in this methodology, and the cross-information base was better in its presentation. In [22] proposes an element extraction calculation for the worldwide scene dependent on the combination of multi- scale highlight maps. With the thought of the feeling engendering between various figures in the picture, the paper proposes a LSTM based calculation for combination the face highlights among various figures. Trials show that the worldwide scene include extraction calculation proposed in this paper has higher exactness than the worldwide scene highlight extraction calculation dependent on standard organization design. In [23], the normal calculation technique is utilized to meld the facial highlights of various figures. In [24], the highlights of various figures is extricated and the middle of these highlights is utilized as the neighborhood face highlight of the entire picture. In [25] use an enormous edge softmax work for discriminative learning. This work centers around worldwide scene's job in the worldwide level feeling acknowledgment while disregard the feeling engendering between the figures in the picture. In [26] centers around separating the nearby and worldwide highlights while ordering the feeling. In any case, this work additionally disregards the feeling engendering issue.

3. Idea of the Proposed Application.

For making the informational collection Microsoft Access is used. Forward looking pictures are isolated from the informational collection. Forward looking face picture extraction is done for reducing the effect of changing establishments on the face affirmation method system proposed. Forward looking face pictures are secured. Provided from the forward looking picture, neighborhood features such as region of mouth, eyes and nose are removed actually. HG-DCT is practiced by all of the above features. The coefficients of HG-DCT of the above regions are

(5)

taken care of. For assessment these coefficients are used. Rate of affirmation procured with the above local features are stood out from rate of affirmation got when HG-DCT practiced by the overall picture. HG-DCT is a precise and strong face acknowledgment framework and utilizing certain standardization methods, its vigor to varieties in facial calculation and enlightenment can be expanded. Face standardization methods are likewise joined in the face acknowledgment application talked about. In particular, a relative change is utilized to address scale, position, and direction changes in faces. Because of this a gigantic enhancements in acknowledgment rates can be accomplished with such standardization. To the forestall issue of brightening varieties, faces in the information bases utilized for the tests are consistently enlightened. That is, certain brightening standardization procedures are utilized to make all appearances have a similar in general dark scale force. An elective way to deal with the HG-DCT is KLT (Karhunen-Loeve change). This change shows design acknowledgment properties that were to a great extent ignored at first in view of the intricacy associated with its calculation. This change delivers an extension of an information picture as far as a bunch of premise pictures or the supposed "Eigen pictures." But HG-DCT is superior to KLT on account of following examinations.

A multifaceted nature connection between the Karhunen-Loeve change (KLT) and the HG-DCT is that, in the application proposed, using HG-DCT, getting ready fundamentally suggests preparing the HG-DCT coefficients of all the informational collection appearances. Of course, using the KLT, getting ready includes handling the reason changing vectors. This infers that the KLT is even highly exorbitant in computation with regards to HG-DCT. Regardless, when the KLT premise vectors was gotten, it very well could have fought enlisting the coefficients of KLT for affirmation is inconsequential. However, the mthod similarly legitimate for the HG-DCT, having extra course of action in this HG-DCT might misuse actual useful estimations of computation.

HG-DCT is a notable sign examination device utilized in pressure because of its minimal portrayal power. It is realized that the KLT is the ideal change as far as data pressing, in any case, its information subordinate nature makes it infeasible to carry out in some down to earth undertakings. In addition, HG-DCT intently approximates the conservative portrayal capacity of the KLT, which makes it a valuable device for signal portrayal both as far as data pressing and as far as computational intricacy because of its information autonomous nature. HG-DCT assists with isolating the picture into parts of contrasting significance (concerning the picture's visual quality). HG-DCT is theoretically like DFT, by means of changing a sign or a picture to the recurrence area from the spatial space.

Figure 1: Transforming image to frequency domain from spatial domain.

(6)

Hybrid graph based Cosine Transform Encoding

The fundamental encoding activity of the HG-DCT is as per the following:

• The size of the data picture is A x B.

• m(a,b) is the power of the pixel at y(a,b).

• M(a,b) represents coefficient of HG-DCT for the y(a,b) pixel.

• For maximum pictures, sign energy’s enormous piece operating at low frequency rates. This shows the HG-DCT upper left corner.

• Compression is developed because of the right lower attributes address higher frequency rate, and are reliably negligible sufficient to pardoned through unimportant apparent mutilation. The coefficients of HG-DCT’s yield pack contains numbers; thin the range of 1024 to 1023.

For calculation, this is less intricate to do furthermore valuable for considering the HG-DCT as a great deal of reason limits that provides a recognized information bunch 8 × 8 size, will be pre- taken care of and dealt with. It joins essentially figuring respects for a convolution cover that get applied in 8 × 8 window.

3.1 Implementation Steps

4 test pictures is contrasted and the data set that comprises of a bunch of pictures of 25 individuals. Four diverse test-techniques are: GLOBAL AND LOCAL, LOCAL, GLOBAL+LOCAL and GLOBAL. In any case, we utilize GLOBAL + LOCAL. In proposed application pictures are sifted utilizing Gaussian low pass channel, honed and changed over to grayscale pictures. At long last force is changed and picture is fragmented and cut to store both worldwide face picture and furthermore its neighborhood highlights, for example, eye nose and mouth independently in the data set after the pictures are standardized.

A. Standardization

Because of pictures of face are taken on the same day or different day with various snapshots, the force for every image will show assortments. Avoiding the light force assortments, normalizing the test pictures with the ordinary power regard to the enrolled picture. The enrolled picture’s ordinary power worth is resolved by adding altogether pixel regards parceled through full scale pixel’s number. Also, test picture’s typical force worth is resolved. The normalization regard is resolved by:

Value of Standardization = Enlisted Image’s average worth / Tried Image’s average worth.

(7)

The increased worth with the test picture’s every pixel. In this manner we get a standardized picture having a normal power regarding that of the enlisted picture.

Next is to perform crisscross filtering and apply HG-DCT.

B. Crisscross examining:

The motivation behind Zigzag Scanning is to:

• Map M x N measurement vector to 1 x N measurements

• Group low recurrence coefficients at the vector’s highest point

Nearby highlights like nose, mouth and eyes are likewise extricated and HG-DCT was practiced by the above highlights. Contingent on acknowledgment rate got for every element, weightage provided by them and afterward joined.

Figure 2: Zigzag scanning

In the wake of removing the highlights, the pictures are thought about utilizing HG-DCT and the correlation is finished by captivating the Euclidean distance among the test and enrolled picture.

Eye-R Eye-L Nose mouth Figure 3: Test image

(8)

For example, venturing through assessment picture’s 50 coefficients and enrolled picture’s 50 coefficients. The Euclidean distance of all of the test picture’s 50 coefficients and the enrolled picture’s50 coefficients were resolved. Sum of 50 coefficient’s Euclidean distance. Leaving the development value alone 'X'. Because of available 15 enrolled pictures, all of the test pictures were differentiated and also 15 selected pictures. Subsequently X1, X2, X3, . . . , X15 regards.

All of these 'X's 15 potential gains are organized by rising solicitation, and having the 'X's base worth is provided in the 'position 1'. Solicitation accompanied is in the provided 'position 2', 'position 3', till 'rank 15'. The 'position 1' picture is seen as the perfect match. In case the 'position 1' picture is comparable to the data picture, then the individual has been seen precisely. Thusly, the rate of affirmation is resolved as the number of pictures extended precisely apparent to the entire number of pictures attempted. The coefficient amount is changed and the rate of affirmation is resolved for all of them utilizing the going with condition:

Acknowledgment rate = Number of accurately perceived picture/Total number of people test picture.

4. Results

Face is related to more exactness and by contrasting the positions for worldwide and neighborhood includes; the bogus acknowledgment rate for HG-DCT is limited. The parametric examination for proposed face acknowledgment framework is talked about in this segment. The evaluation is based on the parameters such as accuracy, precision, recall, and F1 - Score. The stated parameters are evaluated with the estimation of True Positive (TP), False Negative (FN), True Negative (TN), and False Positive (FP).

Accuracy: It defined the number of correctly predicted values to the total number of predictions. It is defined in equation (14)

FN FP TN TP

TN Accuracy TP

 

(12) Recall or Sensitivity: It is defined as the correctly predicted value to the total prediction value. It is defined in equation (15)

FN TP call TP

  Re

(13) Precision: It provides the ratio of true positive values to the total predicted values. It is stated in equation (16)

FP TP ecision TP

  Pr

(14)

(9)

F1 - Score: It provides the ratio between average mean of precision and recall. F1-Score is stated in equation (17)

call ecision

call ecision

Score

F Pr Re

Re

*

*Pr 2

1   (15) The below table-1 shows parametric comparison between existing and proposed technique.

Table 1: Comparison of parameters for existing and proposed method Parameters ANFIS ANS HG-DCT

Accuracy 50 80 95

Precision 80 82 85

Recall 80 85 90

F1-Score 80 83 91

Figure 4: Comparison of Accuracy

(10)

Figure 5: Comparison of Precision

Figure 6: Comparison of Recall

Figure 7: Comparison of F1-Score

(11)

The above figure 4,5,6,7 shows comparison of accuracy, precision, recall and F-1 score for proposed and existing tcehniques.

Table 2: Comparison of parameters for proposed method Parameters HG-DCT

Accuracy 90

Precision 89

Recall 88

F1-Score 87

Figure 8: Overall Comparison of proposed method

The above table-2 shows overall comparison of parameters for proposed technique and figure 8 shows the graphical representation for table 2.

5. Conclusion

When utilizing nearby highlights for acknowledgment, the bogus acknowledgment rate ought to be limited and bogus dismissal rate ought to be augmented when contrasted with that of worldwide highlights. The acknowledgment rate for neighborhood highlights and the acknowledgment rate for worldwide highlights utilizing HG-DCT are determined. Correlation between HG-DCT worldwide highlights and HG-DCT nearby highlights is finished. The

(12)

acknowledgment rate improves when pictures are standardized. At the point when nearby and worldwide highlights are consolidated, HG-DCT gives a generally high acknowledgment rate.

The value of standardization is found with the support of crisscross examination by ranking the pictures in database. The parameters such as accuracy, precision, recall, and F1 – Score are utilized to determine the result of proposed HG-DCT by comparing with two existing methods such as ANFIS and ANF, hence the proposed method achieves 90%of accuracy, 89% precision, 88% recall and 87% of f1-score.

6. Future Enhancement

This technique can be improved by utilizing shading nearby surface highlights for shading face acknowledgment. Extraordinary public face data sets like Yale data set, Olivetti Research Laboratory (ORL) data set and FERRET data set can be utilized to store the images. Further to set up the character of the criminal biometric data and pictures of lawbreakers that can be gotten by any picture obtaining framework, similar to say, CCTV recordings is proposed to be utilized.

References:

[1] M. Singh, R. Singh and A. Ross, "A comprehensive overview of biometric fusion", Inf.

Fusion, vol. 52, pp. 187-205, Dec. 2019.

[2] S. Y. Wong, K. S. Yap, Q. Zhai and X. Li, "Realization of a hybrid locally connected extreme learning machine with DeepID for face verification", IEEE Access, vol. 7, pp.

70447-70460, 2019.

[3] S. Y. Wong, K. S. Yap, Q. Zhai and X. Li, "Realization of a hybrid locally connected extreme learning machine with DeepID for face verification", IEEE Access, vol. 7, pp.

70447-70460, 2019.

[4] A. Sampathkumar, Mulerikkal, J. & Sivaram, M. Glowworm swarm optimization for effectual load balancing and routing strategies in wireless sensor networks. Wireless Networks, vol. 26, no. 6, 4227–4238 (2020).

[5] M. Zhao, Z. Jia and D. Gong, "Improved two-dimensional quaternion principal component analysis", IEEE Access, vol. 7, pp. 79409-79417, 2019.

[6] M. Hu, Y. Zheng, C. Yang, X. Wang, L. He and F. Ren, "Facial expression recognition using fusion features based on center-symmetric local octonary pattern", IEEE Access, vol. 7, pp. 29882-29890, 2019.

[7] A. R. Rivera, J. R. Castillo and O. O. Chae, "Local directional number pattern for face analysis: Face and expression recognition", IEEE Trans. Image Process., vol. 22, no. 5, pp. 1740-1752, May 2013.

(13)

[8] M. O. Oloyede, G. P. Hancke and H. C. Myburgh, "Improving face recognition systems using a new image enhancement technique hybrid features and the convolutional neural network", IEEE Access, vol. 6, pp. 75181-75191, 2018.

[9] Sampathkumar, Rastogi, R., Arukonda, S. Achyut Shankar, Sandeep Kautish & M.

Sivaram “An efficient hybrid methodology for detection of cancer-causing gene using CSC for micro array data” Journal of Ambient Intelligence and Human Comput (2020),Springer. https://doi.org/10.1007/s12652-020-01731-7.

[10] X. Wu and J. Sun, "Joint-scale LBP: A new feature descriptor for texture classification", Vis. Comput., vol. 33, no. 3, pp. 317-329, Dec. 2015.

[11] A. Sampathkumar, Vivekanandan. P “Gene Selection Using PLOA Method In Microarray Data For Cancer Classification” Journal of Medical Imaging and Health Informatics (2019). 9, 1294-1300.

[12] Li, Li, et al. "Face recognition using gabor-based feature extraction and feature space transformation fusion method for single image per person problem." Neural Processing Letters 47.3 (2018): 1197-1217.

[13] Saeed, Usman, Khalid Masood, and Hussain Dawood. "Illumination normalization techniques for makeup-invariant face recognition." Computers & Electrical Engineering 89 (2021): 106921.

[14] Umer, Saiyed, Bibhas Chandra Dhara, and Bhabatosh Chanda. "Face recognition using fusion of feature learning techniques." Measurement 146 (2019): 43-54.

[15] V. Gosavi, A. Deshmane, G. Sable, "Adaptive Neuro Fuzzy Inference System for Facial Recognition," IOSR Journal of Electrical and Electronics Engineering , vol. 14, no. 3, pp.

15-22, 2019.

[16] V Jaiswal, V. Sharma, S. Varma, "An implementation of novel genetic based clustering algorithm for color image segmentation," TELKOMNIKA Telecommunication Computing Electronics and Control, vol. 17, no. 3, pp. 1461-1467, 2019.

[17] J. Li, T. Qiu, C. Wen, K. Xie, F. Wen, "Robust Face Recognition Using the Deep C2D- CNN Model Based on Decision-Level Fusion," Sensors , vol. 18, no. 7, pp. 1-27, 2018.

[18] Wan, Jianwu, Yinjuan Chen, and Bing Bai. "Joint feature extraction and classification in a unified framework for cost-sensitive face recognition." Pattern Recognition 115 (2021):

107927.

(14)

[19] Gupta, Surbhi, Kutub Thakur, and Munish Kumar. "2D-human face recognition using SIFT and SURF descriptors of face’s feature regions." The Visual Computer (2020): 1- 10.

[20] Tripathi, Rajesh Kumar, and Anand Singh Jalal. "Novel local feature extraction for age invariant face recognition." Expert Systems with Applications 175 (2021): 114786.

[21] A. Sampathkumar, P. Vivekanandan “Gene Selection Using Multiple Queen Colonies In Large Scale Machine Learning”in International journal of Electrical engineering,vol.

9,no. 6, pp. 97-111,2019.

[22] Yu, D., Xingyu, L., Shuzhan, D., & Lei, Y. (2019). Group emotion recognition based on global and local features. IEEE Access, 7, 111617-111624.

[23] L. Surace, M. Patacchiola, E. B. Sönmez, W. Spataro and A. Cangelosi, "Emotion recognition in the wild using deep neural networks and Bayesian classifiers", Proc. 19th ACM Int. Conf. Multimodal Interact. (ICMI), pp. 593-597, Nov. 2017.

[24] A. Rassadin, A. Gruzdev and A. Savchenko, "Group-level emotion recognition using transfer learning from face identification", Proc. 19th ACM Int. Conf. Multimodal Interact. (ICMI), pp. 544-548, Nov. 2017.

[25] L. Tan, K. Zhang, K. Wang, X. Zeng, X. Peng and Y. Qiao, "Group emotion recognition with individual facial emotion CNNs and global image based CNNs", Proc. 19th ACM Int. Conf. Multimodal Interact. (ICMI), pp. 549-552, Nov. 2017.

[26] Q. Wei, Y. Zhao, Q. Xu, L. Li, J. He, L. Yu, et al., "A new deep-learning framework for group emotion recognition", Proc. 19th ACM Int. Conf. Multimodal Interact. (ICMI), pp.

587-592, Nov. 2017.

Referințe

DOCUMENTE SIMILARE

The evolution to globalization has been facilitated and amplified by a series of factors: capitals movements arising from the need of covering the external

Using a case study designed for forecasting the educational process in the Petroleum-Gas University, the paper presents the steps that must be followed to realise a Delphi

m3 and a surface of 75 ha which would supply with drinkable water the resorts Azuga, Bu şteni, Sinaia, Comarnic and Breaza used also to produce the electric energy for

These themes are: frequency use of New Information and Communication Technologies (NICT), use of NICT in order to find out information about church, cult

The best performance, considering both the train and test results, was achieved by using GLRLM features for directions {45 ◦ , 90 ◦ , 135 ◦ }, GA feature selection with DT and

Key Words: American Christians, Christian Right, Christian Zionism, US-Israel Relations, Conservative Christians Theology, State of Israel, Jews, Millennial beliefs,

However, the sphere is topologically different from the donut, and from the flat (Euclidean) space.. Classification of two

The number of vacancies for the doctoral field of Medicine, Dental Medicine and Pharmacy for the academic year 2022/2023, financed from the state budget, are distributed to