Volume 5. No. 4. June, 2012

Vol 5. No. 4. June, 2012

ISSN 2006-1781 African Journal of Computing & ICT © 2012 Afr J Comp & ICT – All Rights Reserved www.ajocict.net

Volume 5. No. 4. June, 2012 www.ajocict.net

All Rights Reserved © 2012

Published for the Institute of Electrical & Electronics Engineers (IEEE) Computer Chapter Nigeria Section

By

The Trans-Atlantic Management Consultants 2580 Fairham Drive Baton Rouge LA, USA

i

Vol 5. No. 4. June, 2012


Published Papers 1-8

A Non-Iterative Automated Mechanism for Image Inpainting A.R. Zubair Dept of Electrical Engineering, University of Ibadan, Nigeria

9-18

Complementing for Permanent Identity Theft in Biometrics Using Visual based Challenge Response Systems T. Mapayi, A. Salami & A.B.C Robert Department of Computer Science, Bells University of Technology, Ota, Nigeria

19-32

Determinants of Users' Choice of Mobile Service Providers in the Nigerian Telecommunication Market W. Olatokun & S. Nwonne Africa Regional Centre for Information Science, University of Ibadan, Nigeria

33-42

Context-Aware Stemming Algorithm for Semantically Related Root Words K.K. Agbele, A.O. Adesina, N.A. Azeez & A.P. Abidoye Department of Computer Science, University of the Western Cape, Cape Town, South Africa

43-48

International Evaluation of a Localized Geography Educational Software O.S. Asaolu Department of Systems Engineering, University of Lagos, Nigeria

49-52

Intelligent Agents – Autonomy Issues H. C. Inyiama, S. U. Ufoaroh & V. C. Chijindu Dept. of Electronic Engineering, University of Nigeria, Nsukka, Nigeria.

53-62

Evaluating SMS Parsing Using Automated Testing Software A.O. Adesina, K. K. Agbele, A.P. Abidoye & N.A. Azeez Department of Computer Science, University of the Western Cape, Cape Coast, South Africa

63-68

Comparative and Analysis of Adaptive Protection of Distribution Grids with Distributed Generation S. Mohammadi Department of Electrical Engineering, Kermanshah University of Technology, Iran.

69-72

Effects of VCD) Based Instructions on Students’ Learning Outcomes in an Introductory Technology Class I. M. Longe Department of Curriculum & Instructional Technology, Ogun State University, Ago Iwoye, Nigeria

73-80

Designing and Evaluating Performance in Computer Based Word Game: The Nigerscrab Experience Y. O. Folajimi Department of Informatics & Centre for Computational Intelligence, DeMontfort University, Leicester, United Kingdom

81-93

State-of-the Art: A Comparative Analysis of Ontology Matching Systems O. Iroju, A. Soriyan, I. Gambo & Ikono, R. Department of Computer Science, Adeyemi College of Education, Ondo State, Nigeria.

94-102

Synthesis-Algo: An Inclusive Timetable Generating Algorithm A.O. Egwali & F.A. Imouokhome Department of Computer Science, University of Benin, Benin City, Nigeria

103-108 Forecasting Portfolio Investment Using Data Mining L.A. Adebimpe,, O. Adedara, & O.B. Longe Emmanual Alayande College of Education, Oyo, Nigeria 109-120 A fuzzy-Based Intelligent Traffic Control System for Managing VIP-Induced Chaos at Road Intersections L. A. Akanbi & E. A. Olajubu Obafemi Awolowo University, Ile-Ife, Nigeria 121-126 Framework for a Web-Based Spatial Decision Support System for Academic Advising. E.I. Nwelih & S.C. Chiemeke Department of Computer Science, University of Benin, Benin City, Nigeria Call for Papers

ii

Vol 5. No. 4. June, 2012


Editorial Board Editor-in-Chief Prof. Dele Oluwade Senior Member (IEEE) & Chair IEEE Nigeria – Computer Chapter. Dean - College of Information & Communication Technology Salem University, Lokoja, Nigeria

Editorial Advisory Board Prof. Gloria Chukwudebe - Senior Member & Chairman IEEE Nigeria Section Engr. Tunde Salihu – Senior Member & Former Chairman IEEE Nigeria Section Prof. A.B. Sofoluwe – Vice Chancellor, University of Lagos, Nigeria Prof. Adenike Osofisan - University of Ibadan, Nigeria Prof. Amos David – Universite Nancy2, France Prof. Clement K. Dzidonu – President Accra Institute of Technology, Ghana Prof. Adebayo Adeyemi – Vice Chancellor, Bells University, Nigeria Prof. S.C. Chiemeke – University of Benin, Nigeria Prof. Akaro Ibrahim Mainoma – DVC (Admin) Nasarawa State University, Nigeria Dr. Richard Boateng – University of Ghana, Ghana. Prof. Lynette Kvassny – Pennsylvania State University, USA Prof. C.K. Ayo – Covenant University, Nigeria Dr. Williams Obiozor – Bloomsburg University of Pennsylvania, USA Prof Enoh Tangjong – University of Beau, Cameroon Prof. Sulayman Sowe, United Nations University Institute of Advanced Studies, Japan Dr. John Effah, University of Ghana Business School, Ghana Mr. Colin Thakur - Durban University of Technology, South Africa Mr. Adegoke, M.A. – Bells University of Technology, Ota, Nigeria

Managing/Production Editor Dr. Longe Olumide PhD Southern University and A & M College Baton Rouge, LA, USA Department of Computer Science University of Ibadan, Ibadan, Nigeria

iii

Vol 5. No. 4. June, 2012


Managing Editors Introduction. The ICT strides continues – so are the challenges Events within the last one year showed that Computing and ICT will continue to dominate popular discussions all over the world. What with new advances in mobile developments, tele-medicine, drones for fighting battles and new concerns about viruses that can now close down nuclear plants. We live in a time that can be considered as a watershed of mixed blessings brought about by accelerated developments in computing, ICTs and their applications. The African Journal of Computing & ICT stands at the nexus of providing a platform for contributions to discourses, developments, growth and implementation of Computing and ICT initiatives. We provide a voice for scholars from the developing countries and other nations across the world to contribute to the solution paradigm through timely dissemination of research findings as well as new insights into how to identify and mitigate possible unintended consequences of ICTs. In the last year, we have experienced tremendous impact and acceptance across the globe. This is evidenced by the H-Index of some of our authors ranking as high as 7 based on cited works in the journal. Our impact factor has also grown beyond the 1.7 in 2011 to 2.8 during the first quarter of this year, 2012. In particular is the interest expressed in the Journal by African scholars in Diaspora. Thanks to our authors and reviewers and the Editorial board for continuously striving to provide submissions, insights and comments that has helped improve the quality of the Journal. We have also continued to experience increased interest evidenced by the volume of submissions that we received recently from Asia, America, the United Kingdom and many parts of Africa that are now scheduled to appear in future issues. Surely, the Journal is repositioning itself among the league of impactful Journals in Computing, information systems/science, Information and Communication Technology and other allied fields globally. The African Journal of Computing & ICTs is now indexed in the Cabells Directory of ScientificJourna, Google Scholar, IS Publication Indexing, ScienceCentral.com , Database of Computer Science Journals, Docstoc Database/Indexing and the SCRIBD Research Database. Other indexing schemes are being pursued to further increase access to the journal contents. We remain committed to excellence in publishing and desire that the Afr, J of Computing and ICT be a prime avenue for the dissemination of cutting edge research report by Africans, for Africa and all lovers of research and development all around the world. We will promote indigenous Computing and ICT development through the dissemination of cutting edge research and development report This June 2012 issue of the African Journal of Computing & ICT contain ten (15) articles that have been subjected to rigorous peer review by experts in the subject domain. The articles articulate issues of research and development related to imaging, mobile platform diffusion, algorthims, mathematical modelling, software testing, grid computing, cyber security, e-Learning and multimedia-based pedagogy and games, data mining and knowledge management, fuzzy logic, ontology, intelligent agents and decision support systems. While welcoming you to peruse this Edition of the African Journal of Computing and ICTs, we encourage you to submit your manuscript for consideration in future issues of the African Journal of Computing and ICTs. Have a rewarding reading Thank you Longe Olumide Babatope PhD Managing Editor African Journal of Computing & ICTs

iv

Vol 5. No. 4, June 2012


A Non-Iterative Automated Mechanism for Image Inpainting A.R. Zubair (PhD) Member, IEEE Electrical /Electronic Engineering Department University of Ibadan Ibadan, Nigeria [email protected], [email protected]; +2348023278605

ABSTRACT Portions of an image may be damaged or missing. Inpainting is required to recover the missing portions. Inpainting is cleaning off dirt, filling discolored sports, and repairing torn, warped, or cracked in a damaged image. The existing Navier-Stokes Partial Differential Equations (PDE) method for inpainting is iterative by nature, with a time variable serving as iteration parameter. For reasons of stability, a large number of iterations can be needed which results in a computational complexity that is often too large for interactive image manipulation. A non-iterative automated mechanism for image inpainting is proposed. Colors are treated as fluid that flow or diffuse from the surrounding areas into the empty region. Gains ranging from 9.44 dB to 19.49 dB were obtained with the non-iterative automated inpainting scheme. The automated inpainting scheme overcomes the computational complexity associated with the existing Navier-Stokes PDE inpainting method, and is more suitable for interactive image manipulations. Keyword: Inpainting; Restoration; Image Recovery; Interpolation; Fluid Dynamics.

1. BACKGROUND Inpainting is the restoration of picture's missing pieces. That is cleaning off dirt, filling discolored sports, and repairing torn, warped, or cracked in a damaged image. It is the reconstruction of missing or damaged portions of images [1,2,3,4,11,12,14,19,20]. Portions of an image may be missing or damaged during storage or transmission. Losses of some pixels may be due to mishandling, ageing or deterioration of storage medium. Loss can also occur during transmission; packet loss in wireless transmission for example. Fig. 1 illustrates sampling of one dimensional analog signal and loss of some samples during transmission and/or storage. Pixels of an image are samples of the image which is a multidimensional signal [7,9,13,15,17,18]. Inpainting is also known as retouching. This activity consists of filling in the missing areas or modifying the damaged areas in a non-detectable way by an observer not familiar with the original image. Applications of image inpainting range from restoration of photographs, films and paintings, to removal of occlusions such as text, subtitles, watermarks, stamps and publicity from images. In addition, inpainting can also be used to produce special effects. Fig. 1: Sampling of analog signal and loss of samples. Legends (a) (b) (c) (d) (e)

African Journal of Computing & ICT Reference Format: A.R. Zubair (2012). A Non-Iterative Automated Mechanism for Image Inpainting. Afr J. of Comp & ICTs. Vol 5, No. 4. pp 1-8. © African Journal of Computing & ICT June, 2012 - ISSN 2006-1781

1

Analog One Dimenstional Signal Sampled Version of the Signal Sampled Signal with two adjascent samples missing Sampled Signal with eight adjascent samples missing Sampled Signal with eight non-adjascent samples missing

Vol 5. No. 4, June 2012


Clues from surrounding areas are used to guess what once adorned an image's damaged areas [1,2,10,11]. Visible patterns and structures are then extended into the empty regions. The colors are allowed to diffuse into the missing areas of the image. In general, there is no single correct solution to a given problem. The user has to specify for the computer which areas need to be filled in and precisely what colors, forms, and textures should go into the gaps. One of the automated methods of inpainting is the Navier-Stokes Partial Differential Equations [6,16,20]. The approach uses ideas from classical fluid dynamics to propagate isophote lines continuously from the exterior into the region to be inpainted. Image intensity is seen as a ‘stream function’ for a twodimensional incompressible flow. The method is directly based on the Navier-Stokes equations for fluid dynamics, which has the immediate advantage of well-developed theoretical and numerical results. This approach introduced ideas from computational fluid dynamics into problems in computer vision and image analysis. Fixing digital images is not limited only to filling in blank regions but also includes removal of extraneous objects such as superimposed text, a distracting spectator in the background, or a political foe of the featured person from a given scene. The process could also improve an image's resolution or correct for losses suffered during the transmission of digital images.

    2 255  PSNR = 10 log 10   m n 3 1 2  ∑∑∑ (I i (i, j, t ) − I o (i, j, k ) )    3mn  i =1 j =1 t =1  

2. NON-ITERATIVE AUTOMATED MECHANISM First, the areas of the image to be inpainted are marked by a human agent. The remaining tasks are automated. The image is scanned from left to right; first row to last row. This scanning reveals the location of empty regions. Hence, the empty or missing pixels to be filled are identified. The filling is on first come first serve basis. Recently filled missing pixel can contribute to the next missing pixel. Colors are treated as fluid that flow or diffuse from the surrounding areas into the empty region. Each empty pixel to be filled receives color flows from different directions spanning 165.96o [tan-1(-1/4)] as illustrated in Fig. 2 and Table 1. The directions are limited to less than 180o span to avoid yet-to-be filled missing pixel contributing to the missing pixel currently being filled. This step is necessary in order to avoid propagation of errors. The directions are illustrated in Fig. 2(b) and listed in Table 1.

Damaged old movies or historical films can be repaired. By converting a movie's frames into a sequence of digital images, it's possible to use a computer to detect and repair scratches and dust spots on a given frame by comparing it to adjacent frames and copying image information from intact areas.

* ^

Whether digital or manual, inpainting is always an attempt to make up for lost information. In many situations, there may be multiple solutions to how a gap can be filled to produce a plausible result. Ultimately, judgment resides in the eye of the beholder.

Key

The Navier-Stokes Partial Differential Equations (PDE) method is iterative by nature, with a time variable serving as iteration parameter. For reasons of stability, a large number of iterations can be needed which results in a computational complexity that is often too large for interactive image manipulation [20]. In this work, a non-iterative automated mechanism for image inpainting is proposed. Digital Inpainting is considered from the view point of signal processing [5,20]. The value of peak signal to noise ratio (PSNR) is used to evaluate the error or improvement introduced to an image by any system or process. A higher PSNR indicates improvement or gain. Suppose Ii is the input image into a system or process and Io is the output image from the system or process. The effect of the system or process is described by PSNR given by Eqn. (1) [7,8,9,14,17].

(1)

^ ^

^

Missing Pixel to be filled.

*

Missing Pixel Currently being filled Neighboring Contributing Pixel (npc)

(a) Empty Region and Contributing Pixels 900

900

00

* ^

^ ^

(b) Directions of flow of colors from contributing pixels to the missing pixel currently being filled Fig. 2: Empty Region, Neighboring Contributing Pixels and Directions of Color Flow.

2

Vol 5. No. 4, June 2012


Five pixels along each direction contribute to the missing pixel currently being filled. The contributing pixels along each direction are listed in Table 1. (m,n) is the location of the pixel currently being filled. In Table 1, p is the pixel currently being filled, ncp1 is the neighboring contributing pixel closest to p along the given flow direction. ncp2 to ncp5 are also neighboring contributing pixels.

2.2 Case II: I1 ≥ I2 ≥ I3 ≥ I4 < I5 or I1 < I2 < I3 < I4 > I5 (Continuous trend I1 to I4 only) I5 is neglected. The average of dId and dIc should give dIb if the flow is effective. However, dI = dI d + dI c . The ratio of dIb to bav 2 the average of dId and dIc is given as c = dI b . dIa is taken as a

The pattern of flow along each direction is studied and limiting factor c and weight w are applied to obtain the contribution of the directional flow to the pixel being filled. c and w are determined based on Fig. 3 which shows the neighboring pixels along a direction. Four different cases are identified.

product of c and the average of dId, dIc and dIb as given by Eqn. (5). c(dI d + dI c + dI b ) (5) dI =

2.1 Case I: I1 ≥ I2 ≥ I3 ≥ I4 ≥ I5 or I1 < I2 < I3 < I4 < I5 (Continuous trend from I1 < to I5 ) The average of dIe, dId and dIc should give dIb if the flow is effective. However, dI = dI e + dI d + dI c . The ratio of dIb to bav 3 the average of dIe, dId and dIc is termed limiting factor which is dI b . dIa is taken as a product of c and the average given as c=

dI bav

of dIe, dId, dIc and dIb as given by Eqn. (2).

dI a =

c( dI e + dI d + dI c + dI b ) 4

(2)

If dI a is greater than a constant ck, a weight w is chosen as 1 for the direction concerned and Ip for this direction is set equal to I1 as given by Eqn. (3). Otherwise, w is chosen as 4 and Ip is given by Eqn. (4). If dI a > ck , w = 1 , I p = I1 If dI a ≤ ck , w = 4 , I p = I1 − dI a

ncp5

ncp4

ncp3

ncp2

ncp1

I5

I4

I3

I2

I1

(3) (4)

p

dI bav

a

3

If dI a is greater than ck, w is chosen as 1 for the direction concerned and Ip for this direction is set equal to I1 as given by Eqn. (6). Otherwise, w is chosen as 3 and Ip is given by Eqn. (7). If dI a > ck , w = 1 , I p = I1 If dI a ≤ ck , w = 3 , I p = I1 − dI a

(6) (7)

2.3 Case III: I1 ≥ I2 ≥ I3 < I4 or I1 < I2 < I3 > I4 (Continuous trend I1 to I3 only) I5 and I4 are neglected. dIc should be equal to dIb if the flow is effective. However, this may not be the case. The ratio of dIb to dI b . dIa is taken as a product of c and the average of dIc is c=

dI c

dIc and dIb as given by Eqn. (8). dI a =

c(dI c + dI b ) 2

(8)

If dI a is greater than ck, w is chosen as 1 for the direction concerned and Ip for this direction is set equal to I1 as given by Eqn. (9). Otherwise, w is chosen as 2 and Ip is given by Eqn. (10). If dI a > ck , w = 1 , I p = I1 (9) If dI a ≤ ck , w = 2 , I p = I1 − dI a

(10)

* dIe (I5 - I4 )

dI d (I4 – I 3)

dI c (I3 - I2)

dI b (I2 - I 1)

Ip

2.4 Case IV: I1 ≥ I2 < I3 or I1 < I2 > I3 (Continuous trend I1 to I2 only) I5, I4 and I3 are neglected. w = 1 , I p = I1

dIa (I1 - Ip)

(11)

Fig. 3: Computation of Directional Flow

Ip and w for the twenty directions are determined. Case I indicates the strongest flow and attracts the highest weight of 4 if dI a ≤ ck . Case IV indicates the weakest flow and attracts the

I1, I2, I3, I4, and I5 are the intensities of the five neighboring pixels. Ip is the directional contribution to the intensity of the missing pixel

lowest weight of 1 if dIa ≤ ck. All cases are assigned the lowest weight of 1 if dI a > ck . The intensity of the missing pixel being filled is obtained as

where

θ = tan −1 ( −1 / 4 )

∑ w(θ ) I

I ( m, n) =

p

(θ )

θ = tan −1 ( 0 ) θ = tan −1 ( −1 / 4 )

∑ w(θ )

θ = tan −1 ( 0 )

3

(12)

Vol 5. No. 4, June 2012


Table 1: Neighboring Contributing Pixels along different directions of flow. Pixel Name Directions tan-1(0) tan-1(1/4) tan-1(1/3) tan-1(1/2) p (m,n) (m,n) (m,n) (m,n) ncp1 (m, n-1) (m-1, n-4) (m-1, n-3) (m-1, n-2) ncp2 (m, n-2) (m-2, n-8) (m-2, n-6) (m-2, n-4) ncp3 (m, n-3) (m-3, n-12) (m-3, n-9) (m-3, n-6) ncp4 (m, n-4) (m-4, n-16) (m-4, n-12) (m-4, n-8) ncp5 (m, n-5) (m-5, n-20) (m-5, n-15) (m-5, n-10)

Pixel Name p ncp1 ncp2 ncp3 ncp4 ncp5

tan-1(2) (m,n) (m-2, n-1) (m-4, n-2) (m-6, n-3) (m-8, n-4) (m-10, n-5)



Pixel Name p ncp1 ncp2 ncp3 ncp4 ncp5

tan-1(-3/2) (m,n) (m-3, n+2) (m-6, n+4) (m-9, n+6) (m-12, n+8) (m-15, n+10)

tan-1(-1) (m,n) (m-1, n+1) (m-2, n+2) (m-3, n+3) (m-4, n+4) (m-5, n+5)

Directions tan-1(∞) (m,n) (m-1, n) (m-2, n) (m-3, n) (m-4, n) (m-5, n)

tan-1(2/3) (m,n) (m-2, n-3) (m-4, n-6) (m-6, n-9) (m-8, n-12) (m-10, n-15)


tan-1(3/2) (m,n) (m-3, n-2) (m-6, n-4) (m-9, n-6) (m-12, n-8) (m-15, n-10)




Directions tan-1(-2/3) tan-1(-1/2) (m,n) (m,n) (m-2, n+3) (m-1, n+2) (m-4, n+6) (m-2, n+4) (m-6, n+9) (m-3, n+6) (m-8, n+12) (m-4, n+8) (m-10, n+15) (m-5, n+10)

Depending on the location of the missing pixel being filled, some directions may not be relevant. For example if m