Machine GRAPHICS & VISION, Vol. 13 (2004), No. 1/2:

Banerjee B.:
Recognition of partially occluded shapes using a neural optimization network.
MGV vol. 13, no. 1/2, 2004, pp. 3-23.

The current work presents an algorithm for recognition of partially occluded shapes in a cluttered scene. The images are represented by a sequence of angles subtended at the corner points. The cost due to comparison between the input cluttered scene and the stored images is obtained from a cost function designed to store the obtained information in the form of a cost matrix which is presented to the input of an optimization network. The parameters of the optimization network are determined so as to minimize an energy function, the minima of which occur at the solutions of the problem. The results, as obtained in different domains (2D shapes and projected 3D shapes) with different degrees of occlusion, provide interesting insights into the operation of the algorithm as well as avenues for future research.
Key words: occlusion, angle coding, continuous Hopfield network, cost function, Liapunov energy function.

Han J., Lu Z.:
A cooperative stereo matching and occlusion detection algorithm for stereo coding.
MGV vol. 13, no. 1/2, 2004, pp. 25-38.

This paper presents a novel stereo algorithm for obtaining disparity vectors with simply detected occluded points, which is adaptive to a stereo coding scheme. In this algorithm, we first propose a novel scheme based on an adaptive DT mesh model and an epipolar line constraint principle for stereo matching. Furthermore, a simple occlusion detection algorithm is introduced, which uses two fundamental concepts: the uniqueness assumption and the disparity gradient limit principle. Our technique first extracts some triangular vertices from a reference image using an improved adaptive Delaunay triangulation representation algorithm. As feature points, these vertices are matched in the target image with some correlation measurement. After obtaining the disparity vector of each vertex, the uniqueness assumption and the disparity gradient limit are used to select occluded points and label occluded regions. The disparity vectors of the points among the triangular vertices can be calculated by a six-parameter affine transformation. In terms of advantages for image coding of a DT mesh model, this algorithm is definitely appropriate for stereo coding, which has been proven by experimental results. We have applied this algorithm to some stereo images, and the experimental results show that the veracity of the disparity estimation given in this paper is higher than that of the conventional algorithms. What is more, some occluded points and occlusion regions can be easily detected with this algorithm. In addition, the algorithm can achieve higher coding efficiency as compared to conventional stereo coding algorithms.
Key words: stereo matching, occlusion detection, Delaunay triangulation, disparity vector.

Rodríguez T.:
Adaptive real-time segmentation in traffic scenes.
MGV vol. 13, no. 1/2, 2004, pp. 39-52.

Segmentation is a critical element in any computer vision traffic monitoring system. In this paper a segmentation approach is presented with the aim to obtain accurate, compact movement masks suitable for a subsequent tracking stage. Two different segmentation methods using background subtraction techniques for day and night operation are presented. The methods are designed for all weather, hard real-time operation in outdoor highway multilane scenes observed by a fixed camera.
The key elements presented are: a convenient non-linear image sub-sampling and rectification technique, an adaptive background model estimation and a dynamic adaptation of segmentation parameters.
Key words: adaptive segmentation, motion and scene analysis, traffic surveillance, real-time, image rectification, background model estimation.

Bieda R.:
The thresholding technique in edge detection operators for multichannel images.
MGV vol. 13, no. 1/2, 2004, pp. 53-63.

Edge detection is one of the fundamental operations in computer vision. In the literature there are numerous algorithms for edge detection. This paper presents a new approach to edge detection based on thresholding of a gradient magnitude map. Standard thresholding methods use a single threshold in order to isolate edge points. The introduced technique is based on thresholding with hysteresis. It is obtained by modifying the procedure originally proposed by Canny. The effectiveness of the introduced methods are experimentally verified on standard test images.
Key words: edge detection, non-maximum suppressions, edge thresholding, binarization.

Stapor K., Switonski A.:
Automatic analysis of fundus eye images using mathematical morphology and neural networks for supporting glaucoma diagnosis.
MGV vol. 13, no. 1/2, 2004, pp. 65-78.

In this paper a new method for automatic segmentation and classification of fundus eye images (fei) into normal and glaucomatous ones is proposed. The segmentation of the cup region from the fei makes use of a morphological watershed transformation with markers imposed. New features for quantitative cup evaluation are found based on genetic algorithms with the proposed new fitness function. The computed features are then used in a classification procedure which is based on a multilayer perceptron. The mean sensitivity is 90%, while the mean specificity reaches 85%. The results obtained are encouraging.
Key words: gray-level morphological operators, watershed transformation, genetic algorithms, multilayer perceptron, glaucoma.

Porwik P., Lisowska A.:
The Haar-wavelet transform in digital image processing: its status and achievements.
MGV vol. 13, no. 1/2, 2004, pp. 79-98.

Image processing and analysis based on continuous or discrete image transforms are classic techniques. The image transforms are widely used in image filtering, data description, etc. Nowadays, wavelet theorems make up very popular methods of image processing, denoising and compression. Considering that Haar functions are the simplest wavelets, these forms are used in many methods of discrete image transforms and processing. The image transform theory is a well known area characterized by a precise mathematical background, but in many cases some transforms have particular properties which have not been investigated yet. This paper presents graphic dependences between parts of Haar and wavelets spectra for the first time. It also presents a method of image analysis by means of the wavelet-Haar spectrum. Some properties of the Haar and wavelet spectrum are investigated. Extraction of image features directly from spectral coefficients distribution is presented. The paper shows that two-dimensional products of both Haar and wavelet functions products can be treated as extractors of particular image features. Furthermore, it is also shown that some coefficients from both the spectra are proportional, which simplifies computations and analyses to some degree.
Key words: Wavelets, Haar transform, multiresolution.

Kukharev G., Forczmanski P.:
Data dimensionality reduction for face recognition.
MGV vol. 13, no. 1/2, 2004, pp. 99-121.

In the process of image recognition in most of the applications there is a problem with gathering, processing and storing large amounts of data. A possible solution for reducing this amounts and speeding-up computations is to use some sort of data reduction. Efficient reduction of the stored data without losing any important part of it requires an adaptive method, which works without any supervision. In this article we discuss a few variants of a two-step approach, which involves Karhunen-Loeve Transform (KLT) and Linear Discriminant Analysis (LDA). The KLT gives a good approximation of the input data, however it requires a large number of eigenvalues. The second step reduces data dimensionality further using LDA. The efficiency of KLT depends on the quality and quantity of the input data. In the case when only one image in a class is given as input, its features are not stable in comparison with other images in other classes. In the article we present a few methods for solving this problem, which improve on the ideas presented in [Swets D.L., Weng J.: Using discriminant eigenfeatures for image retrieval. IEEE Trans. PAMI, 18(8), 831-836, 1996; Tsapatsoulis N., Alexopoulos V., Kollias S.: A vector based approximation of KLT and its application to face recognition. Proc. of The IX European Signal Processing Conf. EUSIPCO-98. Island of Rhodos, Greece, 1581-1584, 1998].
Key words: dimensionality reduction, face recognition, eigenvectors, KLT, PCA, LDA.

Doulamis N., Doulamis A.:
Non-linear prediction of rendering workload for grid infrastructure.
MGV vol. 13, no. 1/2, 2004, pp. 123-135.

Grid Computing clusters a wide variety of geographically distributed resources. As a result it can be considered as a promising platform for solving large scale intensive problems. For this reason, it can be viewed as one of the hottest issues in the computer society. A computational intensive application which can be gained from such a Grid infrastructure, is rendering, a process dealing with creating realistic computer-generated image and with many applications ranging from simulation to design and entertainment. To implement, however, a rendering process in a Grid infrastructure prediction of its computational complexity is required. In this paper, this is addressed by using several neural network modules, each of which is appropriate for a given rendering process. For this reason, a feature vector is constructed initially, to describe with high efficiency the parameters affecting the complexity of a rendering algorithm. The feature vector is estimated by parsing a file in a RIB format. Then, prediction is performed using a neural network model. Predictions for three types of rendering algorithms are examined: the Ray Tracing, the Radiosity and the Monte Carlo irradiance analysis.
Key words: Grid computing, 3D rendering, neural networks.

Ballerini L., Bocchi L., Hullberg A.:
Analysis of pores in pig meat images.
MGV vol. 13, no. 1/2, 2004, pp. 137-148.

In this paper we present an image processing application for quantification of pores in pig meat images. The aim is to study the relationship between the number and size distribution of pores and process condition and genotype. We used digital camera photographs of slices of tumbled and non-tumbled processed pork loins of different RN genotype. We describe a method to extract and count the number and sizes of pores. We propose a new approach, based on fractal theory, to quantify the size distribution of pores. Our results show that image analysis in conjunction with fractal models can be useful in this field.
Key words: Pig meat images, pore analysis, pore size distribution, RN genotype, tumbling condition, fractal model.

Cwik A., Rokita P.:
Linking constructive and visual simulations in combat simulators.
MGV vol. 13, no. 1/2, 2004, pp. 149-165.

One of the most important problems in the present military simulators is multi-resolution entity simulation. Up-to-date military combat simulators tend to join constructive and the visual one. The ultimate goal is to give a smooth transition between different organization levels and to fully immerse the commanders in an exercise in a virtual battlefield. This paper deals with the two most important problems resulting from multi-resolution entity simulation - i.e. information disaggregation and discrepancies in different time-scale operations. Selected methods of solving these problems are presented here, as well as our proposal of an inexpensive solution ensuring data consistency.
Key words: constructive simulation, distribute interactive simulation, disaggregation.

Grelowska K., Grabska E., Lachwa A.:
On computer-aided publication design.
MGV vol. 13, no. 1/2, 2004, pp. 167-179.

The article presents a method of computer aided publication design. We consider text as a structure of mark displayed on a certain background, without paying attention to its meaning. Our approach is illustrated by a computer program generating layouts of covers with help of graph grammars.
Key words: computer-aided design, graphs, graph grammars, typesetting.

Hutzler G., Gortais B.:
From computer art to ambient displays.
MGV vol. 13, no. 1/2, 2004, pp. 181-191.

We report in this paper an original experience of a collaboration between a computer scientist, G. Hutzler, and a painter, B. Gortais. This experience started with the development of the computer-art work called the Garden of Chances, which consists in a visual and sonorous representation of the meteorological conditions of a distant place. We then evaluated the efficiency of such a representation to convey useful information and to allow users to make appropriate decisions.
Key words: visualization, multi-agent systems, computer art, ambient displays.


Machine GRAPHICS & VISION, Vol. 13 (2004), No. 3:

From the Publisher

Dear Authors and Readers,

I am sorry to inform you that Nos. 1/2 of Volume 13 (2004) of Machine GRAPHICS & VISION was the last issue edited by Prof. Wojciech S. Mokrzycki, who resigned from the position of Editor-in-Chief of the MG&V journal. As he left our Institute and moved to a University outside Warsaw, he considered it difficult to continue this laborious and time-consuming task. We regret this, since it was him who created this journal and led it for so many years, to bring it to an internationally recognized position among other journals in the domain.

Of course, this could not have happened without the kind help and collaboration of all members of the Advisory Board and Authors. I do hope and expect that they will agree to continue this collaboration, in order to maintain the high quality of the journal.

Consequently, I would also like to inform that starting with this volume, Prof. Juliusz Kulikowski (affiliated at the Institute of Biocybernetics and Biomedical Engineering of the Polish Academy of Sciences, a recognized specialist in Computer Vision) has agreed to replace Prof. W. Mokrzycki as Editor-in-Chief. I am sure that this change will not affect the editorial work on the journal, and any possible drawbacks (like irregularity in publishing the consecutive volumes) will be soon eliminated. The publication procedures remain unchanged.

I am convinced that you will join me in thanking Prof. Wojciech Mokrzycki for his effort and devotion while leading Machine GRAPHICS & VISION, and in a warm welcome to the new Editor-in-Chief, wishing him every possible success in his role.

Piotr Dembinski, Director,
Institute of Computer Science of the Polish Academy of Sciences

Di Stefano L., Marchionni M., Mattoccia S.:
A PC-based real-time stereo vision system.
MGV vol. 13, no. 3, 2004, pp. 197-220.

This paper describes a stereo vision system that enables real-time dense depthmeasurements on a personal computer. The system relies on a very efficient stereo matching engine that, unlike many other approaches which use two distinct matching phases in order to detect unreliable matches, uses a single matching phase. Our matching engine allows for rejecting most unreliable matches by exploiting violations of the uniqueness constraint as well as analysing behaviour of correlation scores. Real-time capability has been achieved by deploying very efficient incremental calculation schemes aimed at avoiding redundant calculations and parallelising the computationally expensive portion of the code with Single Instruction Multiple Data (SIMD) parallel instructions, available nowadays on almost any state-of-the-art general purpose microprocessors. Experimental results on real stereo sequences and preliminary results concerning a 3D people tracking/counting application show the effectiveness of the proposed PC-based stereo vision system for real-time applications.
Key words: stereo, real-time, dense depth map, 3D, SIMD.

Sviták R., Skala V.:
A robust technique for surface reconstruction from orthogonal slices.
MGV vol. 13, no. 3, 2004, pp. 221-233.

The problem of surface reconstruction from sets of planar slices representing cross sections through 3D objects is presented. The final result of surface reconstruction is always based on the correct estimation of the structure of the original object and on the visual quality of the resultant surface. This paper is a case study of the structure reconstruction and the surface formation problems. We present a new approach, which is based on considering mutually orthogonal sets of slices. A new method for surface reconstruction from orthogonal slices is described, and the benefit of orthogonal slices is also discussed. The properties and sample results are presented as well.
Key words: surface reconstruction, orthogonal slices, orthogonal cross-sections, contour correspondence.

Florez-Valencia L., Montagnat J., Orkisz M.:
3D graphical models for vascular-stent pose simulation.
MGV vol. 13, no. 3, 2004, pp. 235-248.

Stents are playing an increasing role in the treatment of arterial stenoses and aneurysms. The goal of this work is to help the clinician in the pre-operative choice of the stent length and diameter. This is done by embedding a model of the stent within a real 3D vascular image. Two models are used. First, a simple geometrical model, composed of a set of circles or polygons stacked along the vessel centerline, is used to simulate the introduction and the deployment of the stent. Second, a simplex-mesh model with an adapted cylindrical constraint is used to represent the stent surface. Another axially constrained simplex-mesh deformable model is used to reconstruct a 3D vessel wall. We simulate the interaction between the vessel wall and the stent by imposing the condition that the model of the vessel locally fits the shape of the deployed stent model. Preliminary quantitative results of the vessel reconstruction accuracy are given.
Key words: vascular pathology, stenosis, stent, 3D image, simulation, simplex mesh, deformable model.

Zaqout I., Zainuddin R., Baba S.:
A feature-based approach for segmenting faces.
MGV vol. 13, no. 3, 2004, pp. 249-259.

Human face detection has always been an important problem for face, expression and gesture recognition. Though numerous attempts have been made to detect and localize faces, these approaches have made assumptions that restrict their extension to more general cases. We identify that the key factor in a generic and robust system is that of using a large amount of image evidence, related and reinforced by model knowledge through a probabilistic framework. In this paper, we propose a feature-based algorithm for segmenting faces that is sufficiently generic and is also easily extensible to cope with more demanding variations of the imaging conditions. The algorithm detects feature points from the image and groups them into face candidates using geometric and grey level constraints. Preliminary results are provided to support the validity of the approach and demonstrate its capability to segment faces under different scales, orientations and viewpoints.
Key words: feature-based model, perceptual grouping, partial face groups, candidate feature map.

Lisowska A.:
Extended wedgelets - geometrical wavelets in efficient image coding.
MGV vol. 13, no. 3, 2004, pp. 261-273.

In the modern world, image coding, and especially image compression, plays a very important role. There are well known and recognized theories concerning this topic, such as, for example, Fourier and wavelets theories. Both of these theories allow for representation of images in a sparse way. Unfortunately wavelets, though very good in catching point discontinuities, cannot properly catch line discontinuities often present in images, that is, edges. As a remedy for this problem, the new theory of geometrical wavelets has arisen. In the paper we present a new and fashionable, hence well known, theory of geometrical wavelets called wedgelets, which allows us to code images with edges in a very efficient way. Moreover, we present the new improvement in the wedgelets theory. This improvement - the theory of extended wedgelets - allows us to represent images in a more sparse and efficient way than in the case of the known wedgelets. Such representation allows us to get a higher compression ratio, together with better visual effects. Furthermore, the application to image coding is also presented. The performance of the improvement has been confirmed both theoretically and experimentally.
Key words: wedgelets, geometrical wavelets, multiresolution, image coding.

Bisoi A.K., Mishra S.N., Mishra J.:
Growing a class of fractals based on combination of classical fractals and recursive mathematical series in L-systems.
MGV vol. 13, no. 3, 2004, pp. 275-288.

L-systems can be used to generate fractals. In this paper L-systems with a standard turtle alphabet are combined with L-systems which emulate simple number sequences. The combination of two such L-systems is achieved by concatenating their axioms as well as merging their rule sets. This can happen in several ways, depending in part on the sequence in which axioms and right-hand sides of rules are concatenated. The purpose of this paper is to combine these two applications of L-systems to generate novel fractals. A turtle interpretation is applied to the yield of the generated hybrid L-systems resulting in new fractal images.
Key words: L-systems, fractals, recursive procedure.

Mavromatis S., Bo&iuml J-M., Bulot R., Sequeira J.:
Texture analysis using directional local extrema.
MGV vol. 13, no. 3, 2004, pp. 289-302.

In this paper, we propose a new formalism that enables to take into account textural features of the image in a very robust and selective way. This approach also permits visualization of these features so experts can efficiently supervise an image segmentation process based on texture analysis. The texture concept has been studied through different approaches. One of them is based on the notion of ordered local extrema and is very promising. Unfortunately, this approach does not take into account texture directionality; and the mathematical morphology formalism, on which it is based, does not enable extensions to this feature. This has led us to design a new formalism for texture representation capable of including directionality features. It produces a representation of texture-relevant features in the form of a surface z = f(x, y). The visualization of this surface gives experts sufficient information for discriminating different textures. We illustrate this approach by a set of results showing its interest in the frame of supervised image segmentation.
Key words: image segmentation, texture analysis, scientific visualization, ordered local extrema, directional texture.

Flasinski M:
Book Review
MGV vol. 13, no. 3, 2004, p. 303.

[Tadeusiewicz R., Ogiela M.R.:
Medical Image Understanding Technology - Artificial Intelligence and Soft-Computing for Image Understanding.
Studies in Fuzziness and Soft Computing, vol. 156, Springer-Verlag 2004]

Machine GRAPHICS & VISION, Vol. 13 (2004), No. 4:

Bhuiyan M.A., Hama H.:
3D reconstruction of parametric curves: recovering the control points.
MGV vol. 13, no. 4, 2004, pp. 307-328.

This article introduces a new curve reconstruction method based on recovering the control points of parametric cubic curves. The method developed here has two stages: finding the 3D control points of parametric curves and reconstruction of free curves. The 3D control points of curves are computed from 2D image sequences by using projective reconstruction of the 3D control points and the bundle adjustment algorithm. The relationships among parametric curves, such as Hermite curves, Bézier curves and B-spline curves, are established so that a curve of any model can be achieved for best fitting. Some experiments are performed to show the performance and effectiveness of the algorithm.
The method is based on the slope following and learning algorithm, which provides an efficient way of finding the 3D control points of any type of cubic Bézier curves. This method, which is an extension of our previous work on recovering control points of 2D Bézier curves, can automatically fit a set of data points with piecewise geometrically continuous cubic parametric curves. The experimental results demonstrate that our method is a fast and efficient way of recovering 3D control points of parametric curves, matching free curves and shape reforming.
Key words: parametric curve, Bézier curve, bundle adjustment, 3D control point, projective reconstruction, Hausdorff distance, apple-node distance.

Patera J., Skala V.:
A comparison of fundamental methods for iso surface extraction.
MGV vol. 13, no. 4, 2004, pp. 329-343.

This paper compares four fundamental methods for iso surface extraction based on cell decomposition to tetrahedra. The methods are compared both on mathematically generated data sets as well as on real data sets. The comparison using mathematical data is made from different points of view such, as area approximation, or volume approximation. On the other hand, the Hausdorff distance and the root mean square are used to compare methods on real data sets. The presented comparison can be of help when deciding which of the tested methods to choose, as well as when we need to compare a newly developed method with other existing approaches.
Key words: comparison, iso-surface extraction, error, Hausdorff distance, volume data, computer graphics.

Valev V.:
Supervised pattern recognition with heterogeneous features.
MGV vol. 13, no. 4, 2004, pp. 345-353.

In this paper, we address the supervised pattern recognition problem with heterogeneous features, where the mathematical model is based on construction of thresholds. Non-Reducible Descriptors (NRDs) for fuzzy features are obtained through the use of a threshold value, which is calculated based on the distance between patterns. In case of solving the problem with real features, the mathematical model for construction of thresholds is based on parallel feature partitioning. Boolean formulas are used to represent NRDs.
Key words:supervised pattern recognition, heterogeneous features, non-reducible descriptors.

Klepaczko A., Materka A.:
Clustering quality based feature selection method.
MGV vol. 13, no. 4, 2004, pp. 355-376.

This paper focuses on the problem of dimensionality reduction for objects described by a large number of features. The emphasis is put on the issues of grouping unlabelled data sets, where information about class-membership of observations is unavailable. Commonly used feature extraction methods for unsupervised classification tasks (such as PCA) are not applicable when information necessary for partitioning of the data set is not represented by the data structure as a whole, but is hidden in a limited number of features only. Thus, we propose a novel technique for choosing the best discriminative data features in an unsupervised manner. Our approach is based on data clustering and on clustering quality measures. The method is straightforward but proved perceptive and efficient. Since the research was primarily motivated by the specific problem of classifying MRI data, performance of the constructed algorithm is studied in application to textured image analysis.
Key words: dimensionality reduction, clustering, quality of clustering, texture analysis.

Kukharev G., Nowosielski A.:
Fast and efficient algorithm for face detection in colour images
MGV vol. 13, no. 4, 2004, pp. 377-399.

Detecting human faces automatically is an indispensable step in any system that exploits information content in a face. The surveillance systems are of particular interest in that field. To correctly recognize a person from an input image, presence and location of the head must be determined in the first step. If the face detection subsystems fails, the performance of the whole system suffers. In the article a scheme for detecting faces in colour images is proposed. It utilizes the skin colour method with a new approach to detecting skin colour pixels in the RGB colour space. The edge information is used to increase the distinction between the skin-coloured face patches and the background. It is followed by a scan line candidate determination algorithm. A new method (combining profiles, geometrical moments, the use of the R-B colour subspace and grey level images) for eyes localization is presented. The verification is finally made based on a well-known template matching approach. After new methods and modifications were evaluated, a system based on the proposed scheme was built and tested.
Key words: face detection, eyes localization, colour segmentation, skin-colour, template matching, real-time.

Ramoser H., Birchbauer J., Bischof H.
Computationally efficient and reliable fingerprint mosaicking on embedded hardware using minutiae.
MGV vol. 13, no. 4, 2004, pp. 401-415.

Fingerprint matching is a common technique for biometric authentication. Solid state sensors allow the use of fingerprint recognition in small sized embedded systems. The size of these sensors makes it necessary to store several impressions of the same finger to provide good coverage of the entire fingertip. In order to reduce memory requirements and matching time all these impressions can be fused into a single larger image. Memory constraints imposed by embedded computers prohibit the use of images. A fingerprint is therefore represented as a set of minutiae coordinates and minutiae angles. We present a two stage approach to combine two fingerprints. First, a RANSAC based method is used to determine a rigid transformation which roughly aligns the two fingerprints. Second, the transformation is optimised using a robust least median of squares solution. The reliability of the method is demonstrated on a large synthetic dataset and real fingerprint images. The computational complexity and memory requirements allow implementation of the algorithm on embedded hardware.
Key words:random sample consensus, point set alignment, embedded system, minutiae alignment, fingerprint stitching.

Przytulska M.:
A method for computer analysis of sequences of cardiac ventriculografical images.
[Dissertation abstract]
MGV vol. 13, no. 4, 2004, p. 417.

Goszczynska H.:
Computer analysis of coronarographic images aimed at coronary blood flow estimation.
[Dissertation abstract]
MGV vol. 13, no. 4, 2004, p. 418.

Reviewers' index

Authors' index

Contents of volume 13, 2004