This article was downloaded by: [Umeå University Library]On: 22 November 2014, At: 19:38Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Intelligent Automation & Soft ComputingPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/tasj20

Auto City: A System for Generating 3D Virtual Cities forSimulation Systems on GIS MapsH. Haldun Göktaş a , Abdullah Çavuşoğlu a & Baha şen a

a Technical Education Faculty , Gazi University , Ankara , TurkeyPublished online: 01 Mar 2013.

To cite this article: H. Haldun Göktaş , Abdullah Çavuşoğlu & Baha şen (2009) Auto City: A System for Generating3D Virtual Cities for Simulation Systems on GIS Maps, Intelligent Automation & Soft Computing, 15:1, 29-39, DOI:10.1080/10798587.2009.10643013

To link to this article: http://dx.doi.org/10.1080/10798587.2009.10643013

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Intelligent Automation and Soft Computing, Vol. 15, No. 1, pp. 29-39, 2009 Copyright © 2009, TSI® Press

Printed in the USA. All rights reserved

29

AUTO CITY: A SYSTEM FOR GENERATING 3D VIRTUAL CITIES FOR

SIMULATION SYSTEMS ON GIS MAPS

H. HALDUN GÖKTAŞ, ABDULLAH ÇAVUŞOĞLU*, BAHA ŞEN

Gazi University Technical Education Faculty

Ankara, Turkey

ABSTRACT—3D Virtual Reality (VR) platforms developed on computers can prevent users from possible risks of real physical environments. We are developing hardware and software based training system, in which training of driver candidates is possible. As a part of the system, virtual city generation is essential. We devised an algorithmic system which automatically generates virtual cities supporting several different map formats (i.e. DTED, HTF, SRTM, and DEM). Depending on the conditions submitted, the system generates different layouts for the cities even on the same geographical site. The current mechanism is based on Height Tile File (HTF) maps and initially The New York city model has been successfully implemented and real time 3D navigation in the city using a standard Intel P4 based PC is possible. Key Words: Simulation Systems, Virtual City Generation, Virtual Reality

1. INTRODUCTION Simulation platforms are among the first systems using computing technologies. Pilots and driver

training and radioactive material handling simulations are among the situations which may be examples of such systems. Therefore, the situation and relevant condition based parameters are often set up –virtually- on a computer enabling us to examine the behaviour of the subjects on the simulation system. In VR systems, the man-machine interaction is generally examined in terms of several aspects being considered. The first step of producing such systems is to produce a database of the virtual environment. In VR systems, generally, the activities are restricted to either a certain artificial or simulated real (i.e. physically existing) place that has been embedded into the simulation system. Therefore the users are generally confined to a rather poor environment with a certain predefined scenario (e.g. such as in the game systems).

The studies focused on VR systems [1, 2] could be used to set up platforms for training people. On such platforms, training is provided with big savings, in terms of cost and security. As the algorithms used to produce virtual cities evolve, such systems shall give us better environment and conditions needed for training people. For example, traffic simulators usually provide a mechanism for the beginner drivers to improve their driving abilities in a virtual environment. In such systems, often, a static driving condition is provided. Here, the word ‘static’ means that, the simulation generally involves a certain region of a city, or surface conditions. In this study, we are giving the details of an automatic city generation algorithm which is a part of our complete system illustrated in Figure 1. The system consists of an automatic city generator; a hydraulic piston based electromechanical mechanism to provide realistic altitude changing effects to the driver candidates, and a mechanism to evaluate the behaviour of the drivers. Currently, the electromechanical platform and the city generator are complete and we are now currently dealing with integration of the rest of the system components together.

To create virtual cities, our system is currently capable of employing a number of map formats including DTED (level 0, 1), DEM (level 0, 1, 2, and 3), SRTM (100 meters) and HTF formats. However,

* Corresponding author Tel.: +90-533-3746924 Fax: +90-312-2120059 E-mail address: acavus@gazi.edu.tr (A.Çavuşoğlu)

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

30 Intelligent Automation and Soft Computing

Rule Base

Automatically Generated Cities

Software and Hardware Based Simulation

Platform

The Evaluation of Drivers Behaviour

User Interaction

Figure 1. The General overview of the driver’s simulator

the higher cost of obtaining high resolution DTED, DEM and SRTM maps has forced us to use the HTF maps in this application. In addition, this is a widely used format, especially in open source applications, and obtaining relevant surface textures is easy when compared to the above methods.

The virtual cities that we develop are generated using an algorithmic mechanism enabling us to generate different city layouts, even in the same geographical region. This is especially attractive for the game developers since it gives the possibility of alternative environments. Local civil engineering rules are assumed in the application. However, depending on the situations these may be altered. The following section provides a brief overview of studies focusing on the problem of virtual city generation. Section 3 gives the details of our system structure which automatically generates virtual cities. Finally, Section 4 concludes this paper.

2. ALGORITHMIC VIRTUAL CITY GENERATION Among the problems of virtual city generating systems, building placements, planning of population

dispersion over the districts, traffic planning, plans for electricity consumption in different regions may be listed. As mentioned above, simulations provide a test bed where one can study such parameters. Some articles focuses on the problems of exporting 3D CAD based models of houses/buildings to VR platforms [3, 4]. In same cases, one to one copy of real buildings are possible, therefore sometimes an existing city may be re-generated totally in an artificial manner. VR applications do have the potential of constructing and creating artificial (i.e. non-existing) buildings. However, such applications have not been matured enough. For example, conversion of the building data into VR environments is often single sided and inadequate, as there are different approaches to build the models [4]. For example, in [5] the authors offer a system where the user is free from grid structures and have the ability to dynamically form the objects. Another study concentrates on the simulation of “micro persons” existing mutually in a virtual environment using parallel processing techniques [6]. Starting from the same concept another group set up a platform to examine the behaviour of drivers and pedestrians within the context of interaction in between these actors [7]. Another group has developed a platform where the behaviour of drivers and pedestrians are examined in terms of the fast management of the database access [8].

Another application we would like to mention here is an internet based project called VC Net [9]. This study is on an application where the civil engineering students of USA and other countries are involved in building virtual cities using a digital database of various forms over the internet. Instead of producing totally artificial cities automatically (i.e. as we do), another group is involved in developing artificial cities with a different approach: consisting of a mechanism called L-system to generate the roads and a Genetic Algorithm (GA) to generate and place the buildings. The starting point of this work is taking the top views of the cities and sorting out the roads and the rest of the city layouts. Following, using the color densities of the top views of the buildings, the same region of the city with the buildings of adjacent heights is built. The result is a regenerated version of the original city on a computing platform [10].

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

Auto City: A System For Generating 3D Virtual Cities for Simulation Systems on GIS Maps 31

3. THE STRUCTURE OF THE SYSTEM

3.1 Approaches to City Planning Table I presents city formats which are either being adopted intentionally or unintentionally. These

formats shape the streets, roads and building blocks etc. The city planners use these approaches to determine the future appearance of developing cities.

Table I. Sample city formats [1].

Model Name Format Name Sample

Simple Unplanned

New York Rectangular

Paris Circular

San Francisco Height Based

Initially we have adopted the New York City model where the city consists of rectangular blocks.

After the determination of city format, the criteria for setting up the city become important. In other words, the size of total land required for the city is determined, and the city population is used as a parameter for solving it. For example, a highly dense city like Hong Kong will require a small landscape while a city like Istanbul –a highly populated but broadly dispersed city- will require a comparatively excessive amount of land to be established. At this point, we employ general civil engineering rules which are set by the government to regularize the shape of developing cities. These are general, principle based rules and not dependent on any city. They basically define, the total percentage of a block which may be occupied by the buildings, the region based regularizations on the size/height of buildings, the distance in between the buildings on the same block, the minimum and maximum distance of the buildings, within the block, with the block borderline etc. The algorithm takes into account these rules, for example, Table II is set up to determine the land requirement for a sample city of 250.000 people. Here, a class tag is used as a determination factor in making decision on overall selection of the placements for various establishments. This is also used by our GA, where each class of land has certain set of objects –which can be placed- on relevant block, appointed for appropriate establishment.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

32 Intelligent Automation and Soft Computing

Table II. Approximate land requirement for a sample city of 250.000 people.

Establishment Type Required Land Size (m2) Class Tag

Parks 2.500.000 6

Health 1.000.000 5

Cultural 625.000 4

Social 375.000 3

Educational 250.000 2

Technical Infrastructure 1.000.000 1

Administrative 1.250.500 0

Total 7.000.000

3.2 Modelling the Cities on The HTF Maps Our system uses the HTF format. This has a simple file header [11], and is easy to texture map. It is

also the most suitable format supported by our graphics engine (i.e. GLScene, developed under Delphi programming Environment using OpenGL) when compared to rest of the formats. HTF Maps can either be generated using (i.e. transforming) maps from the other GIS based formats or some applications [12] directly generates custom maps of this format with various options effecting the appearance of surface landscapes. A sample picture generated using this format is illustrated in Figure 2. While Figure 3 presents a sample city produced by the suggested methodology. As it may be observed, it is rather plain when compared to the state-of-the-art applications. For example in [13] the authors describe a modeling mechanism where the building generation in the famous game SimCity is employed. Comparing the pictures of our system with this reference, suggests that the textures and material library of this application is much richer than our system. However, considering the purpose of our system, this quality is enough for the current application; however, since this system provides a totally automatic city generation mechanism, this may be useful if employed by the state of the art applications. To enhance the picture quality of our scenes, the selected textures and the number of alternative objects with greater detail may be increased.

Figure 2. HTF generated map with the superimposed textures.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

Auto City: A System For Generating 3D Virtual Cities for Simulation Systems on GIS Maps 33

Figure 3. A sample city produced by the system.

3.3 Deterministic City Generation

3.3.1 Block Generation and Placement The method that we wish to present for automatic artificial city generation is rule based and

deterministic. The rules refer to the ordering and placement mechanism of the buildings on the blocks while the determinism refers to the allocation of certain number of blocks for certain establishments, classified by tags, guided by civil engineering laws. In other words, the appearance of the city is somewhat predetermined by them. As it may be observed from Figure 3, this affects the final appearance of the city (i.e. city outskirts may be populated largely by smaller cottage type houses instead of tall buildings).

The overview of our system that has been used for virtual city generation is presented in Figure 4. This application starts with the presentation of a geographical map to the system. Taking into consideration the appropriate rules, the block and population parameters, the total area required for establishing the city is determined (e.g. such as in Table II). The block size is input as a parameter, therefore depending on this parameter the total number of the blocks –determined by the population size- is determined at this stage. Afterwards, the roads are constructed. The regions left in between the roads are further subdivided to describe the location of the buildings, park sites, schools etc. In the following step, the buildings are created as 3ds objects and at the final stage an interpretation operation is made for visualisation.

We have used a priority based mapping mechanism which is used for homogeneously distributing the relevant establishments (i.e. class) to the appropriate districts in the city. In the end, all the class items produced for the city according to Table II will be consumed. The size of the block is parameterised and input by the user, although this also may have been decided by the program (e.g. randomizing the block size between two margins), we decided to leave this to the user for the sake of flexibility. Using the dimensions of the blocks and the population data the total region required for the city is determined. The block size is represented by:

wha bbb *= , (1)

Where ba is the block area (size), bh represents the block height and bw is for the block width. The total area required for the city is calculated by:

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

34 Intelligent Automation and Soft Computing

HTF Maps

Map Interpriter

Population Data Block Size

Civil Engineering Laws

Ordering of the Blocks by Deterministic Method

Placement of Block and Road Borders

Object Generation and Placement on the Blocks

3DS Object

Database

3D Graphics Engine

Figure 4. Rule based deterministic virtual city generation flowchart.

∑ ∑=

=

=6i

0iiareaarea_city (2)

Where the areai represents the relevant area, illustrated in Table II, needed for the required establishments. For each class of establishments the number of blocks that need to be allocated is determined by:

∑ ∑=a

prk bprk

Countblk _ ∑ ∑=a

adm badm

Countblk _ (3)

Where prk stands for parks and adm stands for administration and so forth. Using these values, the total number of blocks is calculated by:

∑ ∑ += prkCountblkCountblk __ ……. ∑+ admCountblk _ (4)

Using the geographical information we try to fit the city into a rectangular area (where appropriate). Therefore, the number of blocks to be fit into the rectangular diameters (i.e. the width, height of the whole land to be occupied by the city) is calculated by:

From, h

wxy b

bper = we get,

xyperCountblk

nobx ∑=_

, nobx

Countblknoby ∑= _

(5)

The calculated values of nobx and noby, which represents the number of blocks on the x axis and y axis of the block matrix respectively, must be integer numbers and if needed they are rounded. In other

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

Auto City: A System For Generating 3D Virtual Cities for Simulation Systems on GIS Maps 35

words, we make sure that the total area where the whole city is placed is rectangular. The calculated total number of blocks (eq.3) and the number of blocks needed to be distributed among the field types (eq.4) are used to build a block list represented by a 2D matrix. By parsing on this list, the tag of each block is marked. This also means that the distribution of blocks over the GIS map. To obtain a non-uniform distribution of the objects over the city blocks, GA [14], which is described in detail in the following section, is employed.

3.3.2 Object Placements on The Blocks Using Genetic Algorithm Object placements on a block are illustrated in Figure 5. The rectangular Bounding Boxes (BB)

represent a sample block and smaller gray boxes (i.e. genes) represent the buildings/trees etc., to be placed on the block. The box diameter is 256x256 pixels. This corresponds to a block of 256x256 meters (i.e. the block size is selectable between 100-256 meters). Figure 5 a-b-c represents the results of the fitting function obtained from the crossover, mutation and reproduction operators of GA. This operation can be expressed as:

))(()1( tpFtp =+ (6)

(a) (b) (c)

Figure 5. Random object placement over a block using GA.

Where the F function is expressed by;

)( onreproductimutationcrossoverF ++→ (7)

Where the operators of the genetic algorithm can be seen, for each object to be placed on the block a gene value need to be calculated:

)__(___ norientatioYcoordXcoordobjectsofnmbrsizegene ++×= (8)

In this formula coord_X and coord_Y represent bitwise x and y coordinate sizes for 256 pixels where 1 pixel corresponds to 1 meter. For example, for the 6 objects illustrated in Figure 5, the gene size can be calculated as:

102)188(6_ =++×=sizegene Bits (9)

Therefore, the total population size is calculated by:

bits)(genes_of_nmbrmax_size_genesize_pop 26112256102 =×=×= (10)

When calculating the gene size, the coordinate values of x and y is both taken as 256 pixels. Therefore, each of these values is represented by 8 bit binary numbers. The fitting value is represented by a word. In the case, where the objects remain outside the BB they are shifted towards, inside the BB using a randomly generated offset coordinates between the limits (i.e. 0-255). The number of iterations required for fitting the objects in the BB essentially depends on the size of the BB and the total number of the objects –and their sizes- to be fitted. In Figure 6 the result of object placement on a sample block using this algorithm is given. The employment of this algorithm makes sure that, the overall appearance of the city is not quite regular when observed from the top.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

36 Intelligent Automation and Soft Computing

After fitting the blocks over the map, the roads are textured. Buildings, trees and other types of elements forming the object database are produced [15, 16] using 3ds format. In other words, these are formed using 3D modeling software that can provide the outputs in this format. Some of the objects of our library are acquired from the internet, yet requiring some modifications, while some of them are created from the scratch. At this point, these objects (e.g. buildings, trees etc.) are grouped to form object subsets, where each subset of this group is associated with a certain type of establishment represented by a class tag.

At the outset, a correspondence tag for each block was given. This data also help us to determine the type of (or class of) buildings/objects that may be placed on that specific block. In other words, each block can have a number of alternative buildings (i.e. randomly selectable from the relevant subset) to be placed on it. Having a rich library of produced objects can help to generate virtual cities which do not look so similar/regular. The

placement of buildings (i.e. which fall into the same subset) onto a block is done according to the chart provided in Figure 7. We keep master copies of the objects that are contained within the library. In the case of requirements of any objects they are instantiated from this library. The following pseudo code describes the object placement mechanism: For each block in the city do { Filter the object lists to sort out relevant objects to form a sub-list; Select randomly from the sub-list of the objects; Create proxy objects from the master; Get block coordinates; While check_empty_areas(blockList[i]) do { Calculate object coordinates and create a dummy cube for it; Set the properties of the dummyCube; /*texture, light, etc.*/ Place the dummy cube onto the block; }};

The usage of dummy cubes instead of the objects themselves is due to the higher memory consumption which dramatically degrades the graphics performance of the system. The GA works with a sequence of dummy cubes to be placed on the block, afterwards the real object coordinates are transferred onto the block, which in turn speeds up the placement process. In Figure 8 a screen shot of the resulting scene is illustrated.

4. RESULTS AND CONCLUSIONS In this study an automatic algorithmic city generation mechanism has been developed as a part of a

driver’s simulation system. The system is based on the HTF format which easily provides higher resolution real time picture sequences on a standard Intel P4 based system equipped with Nvidia 256MB graphics card with FX 5500 chipset. Figure 9 shows the performance of the mechanism used for object placement. The required processing time logarithmically increases as the number of objects to be processed increases. The proposed system consists of: a digital geographical map, city planning rules which have been embedded into the system and user interaction to input parameters affecting the size and the final appearance of the virtual city. With this approach, we provide a fast and flexible mechanism to automatically generate virtual cities. In the system the constituent parts of a city are predetermined into 7 sub-classes of buildings and objects, providing randomly selectable alternatives for each block. To achieve this, we decided to utilize the widely used 3ds format which enabled us to have the possibility of several ready made, though needing a normalization operation to fit into our system, objects. We keep growing our library with the introduction of new objects.

Figure 6. 3D Object placement on a block using genetic algorithm.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

Auto City: A System For Generating 3D Virtual Cities for Simulation Systems on GIS Maps 37

Object Generator

3DS Object Database

3DS Object Subset Block List

Object Selector (from the subset of 3DS Objects)

Master 3DS Object Texture Material Library

Master 3DS Object

Proxy Object for Master 3DS Object

Add new node

Update Proxy Object Nodes Coordinates and Locate Nodes Using the Genetic Algorithm

3D Graphics Engine

Master 3DS Object Creation Progress

Figure 7. The flowchart of object placement on the surface.

Figure 8. Placing building objects on the blocks.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

38 Intelligent Automation and Soft Computing

Figure 9. System’s object generation performance.

As described above, initially the New York model has been used for city generation. This was mainly due to its straightforwardness and being the easiest model which could be implemented. The city generation mechanism essentially works on blocks, and hierarchical ordering of the blocks produces the final virtual city. Each block is located on a data structure containing its physical coordinates. The sequences of rules arrange/describe the ordering of the buildings to be placed on each block. This corresponds to the buildings, the distance in between them, their heights, the population that they accommodate etc.

The developed mechanism currently produces city layouts as viewed from the sample pictures illustrated by Figures 3 and 8. The main idea which distinguishes the developed system from the others is that, it is almost fully automatic and several different city layouts may be produced with little user interaction. However, there are limitations of the system as well, for example for the time being, the picture quality of the produced database may be regarded rather poor, when compared to the state-of-the-art applications, in addition the system cannot deal with the altitude differences over the surface (common problems of such systems). Also, changing the city format will require more user intervention reducing the automatic nature of the system, for example: with the Paris model the system can achieve equally successful performance, however, with the height based and unplanned city models, working with the suggested mechanism may not be as easy as thought. At the moment we are dealing with two issues: namely, a placement mechanism for the traffic signboards, and different building distribution mechanisms (i.e. on the blocks) based on population density over the districts of the city. Finally, an improvement may be providing the ability to make alterations on the final product with extra user interaction.

REFERENCES 1. C.A. Ellis, S.J. Gibbs, and G.L. Rein, “Groupware - Some Issues and Experiences”,

Communications of the ACM, 34(1): 38-58, 1991. 2. J. Whyte, N. Bouchlaghem, A. Thorpe, and R. McCaffer, “From CAD to virtual reality: Modeling

approaches, data exchange and interactive 3D building design tools”, Elsevier - Automation in Construction, 10 (1): 43-45, 2000.

3. R. Carey, “The Virtual Reality Modeling Language Explained,” IEEE Multimedia, 05 (3): 84-93, 1998.

4. M. Batty, and B. Jiang, “Multi-Agent Simulation: New Approaches To Exploring Space-Time Dynamics Within GIS”, Working Paper Series of the Centre for Advanced Spatial Analysis (CASA), University College London, 10, 1999.

5. D. Thalmann, N. Magnenat-Thalmann, and S. Donikian, “Half Day Tutorial: Automatic generation of Animated Population in Virtual Environments”, In Eurographics’2004 25th Annual Conference of the European Association for Computer Graphics, Grenoble, 1-5, 2004.

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014

Auto City: A System For Generating 3D Virtual Cities for Simulation Systems on GIS Maps 39

6. F.V. Deriggi, Jr., M.M. Kubo, A.C. Sementille, J.R.F. Brega, S.G. dos Santos, and C. Kirner, “CORBA platform as Support for Distributed Virtual Environments”, Proceedings of IEEE Virtual Reality, 8-13, 1999.

7. W. Fuchs, and A. Martinico, “The VC Net A Digital Study in Architecture”, Elsevier Automation in Construction, 6 (4): 335- 339, 1997.

8. S. Donikian, “VUEMS: A Virtual Urban Environment Modeling System”, In Computer Graphics International’97 IEEE Computer Society Press, Hasselt-Diepenbeek, Belgium, 84-92, 1997.

9. Y. Parish, and P. Muller, “Procedural Modeling of Cities”, ACM Proceedings of Siggraph 2001, Los Angeles, 301-308, 2001.

10. T. Fujii, K. Imamura, T. Yasuda, S. Yokoi, and J. Torikawi, “A Virtual Scene Simulation System for City Planning”, Computer Graphics Developments in Virtual Environments, Academic Press, 6: 483-496, 1995.

11. Internet: GLScene: OpenGL Solution for Delphi, HTF File Format. www.glscene.org 12. Internet: L3DT – Large 3D Terrain Generator, File formats page. http://users.tpg.com.au/blakest2/

l3dt/formats.htm 13. T. Lechner, B. Watson, P. Ren, U. Wilensky, S. Tisue, and M. Felsen, “Procedural Modeling of

Land Use in Cities”, Northwestern University Computer Science, Technical Report, NWU-CS-04-38, 2004.

14. A.S. Austin, “An Introduction to Genetic Algorithm”, AI Expert, 48-53, 1990. 15. S. Koç, and U. Çevik, “A New Approach for the Voxelization of Volumetric CSG Graphs”,

Computers & Electrical Engineering, 30(4), 245-255, 2004. 16. R. Lev, G. Elber, and D. Gordon, “SWAPART: Synthetic Object Creation by Part Substitution”, 5th

Israel-Korea Bi-national Conference on Geometric Modeling and Computer Graphics, 103-107, 2004.

ABOUT THE AUTHORS A. Çavuşoğlu graduated from Gazi University in 1985 and received his M.Sc. degree from the same university in 1988. He has received his D. Phil degree from University of Sussex in 1993 from the School of Engineering. He is currently an Associate Professor at Gazi University, Ankara Turkey. His areas of interest include neural network applications, computer graphics and algorithms

H. Haldun Göktas received the B.S. and M.S. degrees from the Department of Electrical and Electronics Engineering from Hacettepe University, Turkey, in 1987 and 1990 respectively, and received Ph.D. degree from the Department of Electronics Engineering, Erciyes University, Turkey, in 1996. In 1996, he joined the Faculty of Department of Electronics and Computer Education, Gazi University, where he is now an Assistant Professor. His current research interests lay in computer graphics and optical fiber communication systems.

B. ŞEN received his B.Sc. in Computer Science Department from the Gazi University, Ankara/Turkey in 1996. He received his M.Sc. degree from Institute of Science and Technology, Gazi University in 1999, and his Ph.D. degree from same department. His research interests include graphics, vision, genetic algorithms, expert systems, geographical information systems, 3d modelling and simulation systems

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

19:

38 2

2 N

ovem

ber

2014