Search

The least-squares fitting minimizes the squares sum of error-of-fit in predefined measures. By the geometric fitting, the error distances are defined as the orthogonal, or shortest distances from the given points to the geometric model to be fitted. Then the nonlinear optimization algorithm can be used to obtain the optimum solution. In this paper, we propose a geometric fitting algorithm for the deformable superquadric model, which is the computation of a measure of vector from each given point to orthogonal contacting point on the superquadric model, and estimates the optimum parameters of the model to minimize the squares sum of error distances. The estimated parameters by the proposed algorithm are invariant to coordinate transformation and we can easily find a physical interpretation of the fitting parameters. [Copyright &y& Elsevier]

In this paper, we present a new motion estimation algorithm that provides accurate optical flow computation under non-uniform brightness variations. The proposed algorithm is based on a regularization formulation that minimizes a combination of a modified data constraint energy and a smoothness measure all over the image domain. The data constraint is derived from the conservation of the Laplacian-of-Gaussian (LoG) filtered image function, which alleviates the problem with the traditional brightness constancy assumption under non-uniform illumination variations. The resulting energy minimization is accomplished by an incomplete Cholesky preconditioned conjugate gradient algorithm. Comparisons of experimental results on benchmarking image sequences by using the proposed algorithm and some of the best existing methods are given to demonstrate its superior performance. [Copyright &y& Elsevier]

Based on the study of winner-update search (WUS) algorithm, a modified WUS (MWUS) algorithm for fast block-matching motion estimation is proposed in this paper. The computational reduction without any degradation in predicted image comes from fast elimination of impossible motion vectors. We obtain faster elimination of inappropriate motion vectors using the efficient lower bounds set computed according to the edge feature of predicted image. Experimentally, we remove the unnecessary computations by 93.5–98.5% with our proposed MWUS algorithm compared with the FS algorithms. Experimental results also indicate that our algorithm is better than the other winner-update strategy based algorithm. [Copyright &y& Elsevier]

In this paper, an empirically optimized channel-matched quantizer, and a joint stochastic-control based rate controller and channel estimator for H.261 based video transmission over a noisy channel is proposed. The rate controller adaptively learns to choose the correct channel matched quantizer using a stochastic learning algorithm. The stochastic automaton based learning algorithm aids in estimating the channel bit error rate based on a one bit feedback from the decoder. The algorithm is observed to converge to the optimal choice of the quantizer very quickly for various channel bit error probabilities and for different video sequences. When compared to traditional channel estimation schemes the proposed technique has several advantages. First, the proposed method results in a significant reduction in the delay and bandwidth requirement for channel estimation when compared to pilot symbol aided channel estimation schemes. Next, the stochastic learning algorithm used to estimate the channel bit error rate has simple computations. This makes it attractive for low power applications such as wireless video communications. This is in contrast to traditional blind channel estimation schemes that are computationally expensive, in general. [Copyright &y& Elsevier]

Because the estimation of a fundamental matrix is much dependent on the correspondence, it is important to select a proper inlier set that represents variation of the image due to camera motion. Previous studies showed that a more precise fundamental matrix can be obtained if the evenly distributed points are selected. When the inliers are detected, however, no previous methods have taken into account their distribution. This paper presents two novel approaches to estimate the fundamental matrix by considering the inlier distribution. The proposed algorithms divide an entire image into several sub-regions, and then examine the number of the inliers in each sub-region and the area of each region. In our method, the standard deviations are used as quantitative measures to select a proper inlier set. The simulation results on synthetic and real images show that our consideration of the inlier distribution can achieve a more precise estimation of the fundamental matrix. [Copyright &y& Elsevier]

It is widely recognized that the boosting methodology provides superior results for classification problems. In this paper, we propose the boost-clustering algorithm which constitutes a novel clustering methodology that exploits the general principles of boosting in order to provide a consistent partitioning of a dataset. The boost-clustering algorithm is a multi-clustering method. At each boosting iteration, a new training set is created using weighted random sampling from the original dataset and a simple clustering algorithm (e.g. k-means) is applied to provide a new data partitioning. The final clustering solution is produced by aggregating the multiple clustering results through weighted voting. Experiments on both artificial and real-world data sets indicate that boost-clustering provides solutions of improved quality. [Copyright &y& Elsevier]

In this paper we present a new explicit numerical scheme to approximate the solution of the linear diffusion filtering. This scheme is fast, stable, easy to program, applicable to arbitrary dimensions, and preserves the discontinuities of the objects. Experimental results support the efficiency of the proposed approach for the multi-scale detection of edges in greyscale, and color images. [Copyright &y& Elsevier]

Usual approaches for constructing topological maps on discrete structures are based on cellular complexes topology. This paper aims to construct a coherent topological map defined on a square grid from a watershed transformation. The main idea behind the proposed approach is to impose some constraints on the original image in order to obtain good properties of the resulting watershed. We propose a definition of well-composed grayscale images based on the well-composed set theory and the cross-section topology. Properties of two different thinning algorithms are then studied and we show how to obtain a thin crest network. We derive an efficient algorithm that permits the construction of a meaningful topological map, resulting in a topological segmentation, i.e. a segmentation that describes in a coherent framework faces and contours. Finally, we demonstrate the usefulness of this algorithm for multilevel image segmentation. [Copyright &y& Elsevier]

The paper reports on the robust pattern classification of experimental data using a combined approach of symbolization followed by support vector machine (SVM) classification. Symbolization of data removes unwanted features such as noise whereas SVM provides the classification. The SVM parameters are tuned on-line using a genetic–quasi-Newton algorithm. Benchmark examples illustrate the proposed approach. [Copyright &y& Elsevier]

In this paper, an efficient hierarchical clustering algorithm, suitable for large data sets is proposed for effective clustering and prototype selection for pattern classification. It is another simple and efficient technique which uses incremental clustering principles to generate a hierarchical structure for finding the subgroups/subclusters within each cluster. As an example, a two level clustering algorithm––‘Leaders–Subleaders’, an extension of the leader algorithm is presented. Classification accuracy (CA) obtained using the representatives generated by the Leaders–Subleaders method is found to be better than that of using leaders as representatives. Even if more number of prototypes are generated, classification time is less as only a part of the hierarchical structure is searched. [Copyright &y& Elsevier]

Vector quantization (VQ) is an asymmetric coding method and the winner search in encoding process is extremely time-consuming. This property of VQ constrains its practical applications very much. Based on the sum pyramid data structure of a vector, a fast encoding algorithm with PSNR equivalent to full search is proposed in this paper to improve the results that have been given in [Pattern Recognition Lett. 22(3/4) (2001) 373].Three major modifications are made. Because sum can be computed accurately using integer operation, mismatched winners due to roundoff error of taking the mean can be overcome firstly by using sum instead of the mean. Secondly, since sum is computed by 2-pixel-merging other than 4-pixel-merging way, more levels can be constructed in a sum pyramid so that the rejection to current codeword can be realized easier. Finally, codebook is off-line rearranged by sorted real sums directly, and the lower and upper bound for promising codeword class are updated dynamically to narrow search scope further once a better-matched codeword has been found during winner search process.For the same four 512 × 512, 256-level standard test images, experimental results show that the proposed algorithm outperforms the method used in [Pattern Recognition Lett. 22(3/4) (2001) 373] in the meaning of total equivalent distance computation being reduced obviously. [Copyright &y& Elsevier]

A face recognition algorithm based on modular PCA approach is presented in this paper. The proposed algorithm when compared with conventional PCA algorithm has an improved recognition rate for face images with large variations in lighting direction and facial expression. In the proposed technique, the face images are divided into smaller sub-images and the PCA approach is applied to each of these sub-images. Since some of the local facial features of an individual do not vary even when the pose, lighting direction and facial expression vary, we expect the proposed method to be able to cope with these variations. The accuracy of the conventional PCA method and modular PCA method are evaluated under the conditions of varying expression, illumination and pose using standard face databases. [Copyright &y& Elsevier]

In this paper, a nonrigid registration method is presented for accommodating local shape variation when matching monomodal images or multimodal images. Affine transformation is adopted in global registration while local deformation is described by a free-form deformation based on a linear singular blending (LSB) B-spline, which can enhance the shape-control capability of the B-spline. This capability is derived from the blending parameters defined at the B-spline control points. The local deformation of images can be controlled by changing positions of the control points or altering values of blending parameters. Two voxel-based similarity measures, the SSD for monomodal images and mutual information for multimodal images are used. Experiments on simulated and real data have been performed to evaluate the accuracy of our method. We have also compared the results of the proposed approach to those obtained by affine and B-spline registration techniques. The results indicate that the proposed method is much better to describe the deformation than the affine algorithm and B-spline techniques. [Copyright &y& Elsevier]

The recording of interval data has become a common practice with the recent advances in database technologies. This paper introduces clustering methods for interval data based on the dynamic cluster algorithm. Two methods are considered: one with adaptive distances and the other without. [Copyright &y& Elsevier]

In this paper, we propose a method of content-based image classification using a neural network. The images for classification are object images that can be divided into foreground and background. To deal with the object images efficiently, in the preprocessing step we extract the object region using a region segmentation technique. Features for the classification are shape-based texture features extracted from wavelet-transformed images. The neural network classifier is constructed for the features using the back-propagation learning algorithm. Among the various texture features, the diagonal moment was the most effective. A test with 300 training data and 300 test data composed of 10 images from each of 30 classes shows classification rates of 81.7% and 76.7% correct, respectively. [Copyright &y& Elsevier]

As we know, histogram matching is a commonly-adopted technique in the applications of pattern recognition. The matching of two patterns can be accomplished by matching their corresponding histograms. In general, the number of features and the resolution of each feature will determine the size of histogram. The more the number of features and the higher the resolution of each feature, the stronger the discrimination capability of histogram will be. Unfortunately, the increase of histogram size will lead to the decrease of the efficiency of histogram matching because traditional algorithms in evaluating similarity are all relevant to the histogram size. In this paper, a novel histogram-matching algorithm is proposed whose efficiency is irrelevant to the histogram size. The proposed algorithm can be applied to commonly-adopted histogram similarity measurement functions, such as histogram intersection function, L1 norm, L2 norm, χ2 test and so on. By adopting our proposed algorithm, future researchers can focus more on the selection and combination of histogram features and freely adjust the resolution of each feature without worrying the decrease of retrieval efficiency. [Copyright &y& Elsevier]

A modular approach for improving the performance of 2-D Hopfield neural network is presented in this paper. The approach is inspired by biological visual perception phenomenon. The training method introduces a decaying distance factor into the Hebbian learning rule for image processing applications. The value of the distance factor varies based on the spatial location of the neurons with respect to the neuron under consideration. Experiments performed with character images show that the new approach can learn and recognize patterns very effectively. [Copyright &y& Elsevier]

In this paper we have presented a new genetically guided algorithm for solving the clustering problem. The proposed Genetic Clustering Algorithm is basically a two-phase process. At the first phase the original data set is decomposed into a number of fragmented clusters in order to spread the GA search process at the latter phase over the entire space. At the second phase Hierarchical Cluster Merging Algorithm (HCMA) is used. The HCMA is an iterative genetic algorithm based approach that combines some of the fragmented clusters into complete k-cluster. The algorithm contains another component called Adjacent Cluster Checking Algorithm (ACCA). This technique is used for testing adjacency of two segmented clusters so that they can be merged into one cluster. The performance of the algorithm has been demonstrated on several data sets consisting of multiple clusters and it is compared with some well-known clustering methods. [Copyright &y& Elsevier]

A novel contour-based 3D terrain matching method is presented in this paper. In the method, iso-elevation contour map, a compact feature-based representation, is proposed to represent the reference digital elevation model (DEM) and recovered DEM (REM) from real-time data to convert 3D terrain matching to contour-based matching. In the contour-based matching, a normalized wavelet descriptor, which is invariant to 2D rigid transformation and insensitive to noise, is employed to describe contours. A very fast contour-matching algorithm based on normalized wavelet descriptor is presented and the computation cost is analyzed. Control points based on matched contour pairs are located and consistency checking is used to remove the possible mismatching points. 2D rigid image transformation model and parameter estimation using the rule of least root mean square error are discussed. The proposed matching method is robust and effective in computation, and can achieve high location accuracy. All these are demonstrated by experimental results using real terrain data. [Copyright &y& Elsevier]

The multiscale medial axis (MMA) is a robust shape representation in analysing objects in greylevel images. It is computed as ridges in the medialness scale-space of an image. In this paper the multiscale medial properties of overlapping objects are investigated, the result of which is novel to the community in that the existing studies only concentrated on isolated objects. To improve the distinguishability of these objects, a sliding window algorithm is proposed to detect locally optimal scale ridges in scale space. When used in the medialness scale-space of an image, this efficient method can extract a complete set of MMAs for interfering objects, in contrast with the globally optimal scale ridge definition which has been widely used. This is demonstrated in the applications in medical images. [Copyright &y& Elsevier]

In this paper, a simple and robust algorithm is proposed for detecting straight line segments in an edge image. The proposed algorithm is based on small eigenvalue analysis. The statistical and geometrical properties of the small eigenvalue of the covariance matrix of a set of edge pixels over a connected region of support are explored for the purpose of straight line detection. The approach scans an input edge image from the top left corner to the bottom right corner with a moving mask of size k×k for some odd integer k>1. At every stage, the small eigenvalue of the covariance matrix of the edge pixels covered by the mask and connected to the center pixel of the mask is computed. These pixels are said to be linear edge pixels if the computed small eigenvalue is less than a pre-defined threshold value. Several experiments have been conducted on various images with considerable background noise and also with significant edge point location errors to reveal the efficacy of the proposed model. The results of the experiments emphasize that the proposed model outperforms other models specifically the Hough transform and its variants in addition to being robust to image transformations such as rotation, scaling and translation. [Copyright &y& Elsevier]

This paper presents a statistical test and algorithms for patterns extraction and supervised classification of sequential data. First it defines the notion of prediction suffix tree (PST). This type of tree can be used to efficiently describe variable order chain. It performs better than the Markov chain of order L and at a lower storage cost. We propose an improvement of this model, based on a statistical test. This test enables us to control the risk of encountering different patterns in the model of the sequence to classify and in the model of its class. Applications to biological sequences are presented to illustrate this procedure. We compare the results obtained with different models (Markov chain of order L, Variable order model and the statistical test, with or without smoothing). We set out to show how the choice of the parameters of the models influences performance in these applications. Obviously these algorithms can be used in other fields in which the data are naturally ordered. [Copyright &y& Elsevier]

Vision-based systems for finding road lanes have to operate robustly under a wide variety of environmental conditions including large amount of scene clutters. This paper presents a method to find the lane boundaries by combining a local line extraction method and dynamic programming. The line extractor obtains an initial position of road lane boundaries from the noisy edge fragments. Then, dynamic programming improves the initial approximation to an accurate configuration of lane boundaries. Input image frame is divided into sub-regions along the vertical direction. The local line extractor extracts candidate lines of road lanes in the sub-region. Most prominent lines are found among candidate lines by dynamic programming that minimizes the functional which measures the deviation from a virtual straight line. The search framework based on DP method reduces computational cost. Experimental results using images of real road scenes demonstrate the feasibility of the proposed algorithm. [Copyright &y& Elsevier]

This paper addresses a registration refinement problem and presents an accurate and fast point-to-(tangent) plane technique. Point-to-plane approach is known to be very accurate for registration refinement of partial 3D surfaces. However, the computation complexity for finding the intersection point on a destination surface from a source control point is hindering the algorithm from real-time applications. We introduce a novel point-to-plane registration technique by combining the high-speed advantage of point-to-projection technique. In order to find the intersection point fast and accurately, we forward-project the source point to the destination surface and reproject the projection point to the normal vector of the source point. We show that iterative projections of the projected destination point to the normal vector converge to the intersection point. By assuming the destination surface to be a monotonic function in a new 2D coordinate system, we show contraction mapping properties of our iterative projection technique. Experimental results for several objects are presented for both pair-wise and multi-view registrations. [Copyright &y& Elsevier]

This paper presents a generalized version of the conventional k-means clustering algorithm [Proceedings of 5th Berkeley Symposium on Mathematical Statistics and Probability, 1, University of California Press, Berkeley, 1967, p. 281]. Not only is this new one applicable to ellipse-shaped data clusters without dead-unit problem, but also performs correct clustering without pre-assigning the exact cluster number. We qualitatively analyze its underlying mechanism, and show its outstanding performance through the experiments. [Copyright &y& Elsevier]

This paper presents a floating search approach for learning the network structure of Bayesian network classifiers. A Bayesian network classifier is used which in combination with the search algorithm allows simultaneous feature selection and determination of the structure of the classifier.The introduced search algorithm enables conditional exclusions of previously added attributes and/or arcs from the network classifier. Hence, this algorithm is able to correct the network structure by removing attributes and/or arcs between the nodes if they become superfluous at a later stage of the search. Classification results of selective unrestricted Bayesian network classifiers are compared to naı¨ve Bayes classifiers and tree augmented naı¨ve Bayes classifiers. Experiments on different data sets show that selective unrestricted Bayesian network classifiers achieve a better classification accuracy estimate in two domains compared to tree augmented naı¨ve Bayes classifiers, whereby in the remaining domains the performance is similar. However, the achieved network structure of selective unrestricted Bayesian network classifiers is simpler. [Copyright &y& Elsevier]

The objective of this paper is to develop an approach for efficiently and quantitatively evaluating thresholding algorithms for change detection in a surveillance environment. Previous evaluation in the literature has either been subjective or small scale, in part due to the difficulties and/or the time and effort involved in determining appropriate ground truth. In comparison, our automated approach enables us to carry out a more thorough evaluation, and we test the performance of eight different thresholding algorithms using more than 4000 images with two different texture environments. [Copyright &y& Elsevier]

This paper presents a novel algorithm for colour quantization based on frequency diffusion in the histogram of an image. By manipulating a diffusion filter, a cross-space operation is achieved, obtaining quantization effect in a colour space without transforming the image to that space. Results show that the algorithm is effective. [Copyright &y& Elsevier]

Video segmentation is the first step of creating video indices for a video retrieval system. A segmentation algorithm is used to identify shots from video data. In this paper, we propose a histogram-based moment-preserving (HBMP) clustering algorithm for segmenting video data. This algorithm is a hybrid of the shot change detection approach and the clustering approach. The computational results indicate that the proposed algorithm is both effective and efficient with respect to various types of video sequence. [Copyright &y& Elsevier]

In this paper, threshold selection is considered in the continuous image rather than in digital image. We prove that, for each given object within 2D image, its optimal threshold is determined by the mean of the gray values of the points lying on its continuous boundary. Thus, we try to deduce threshold from the gray values of the boundary rather from the gray values of the given discrete sampling points (pixels or edge pixels). By the scheme, we well overcome some disadvantages existing in the threshold methods based on the histogram of edge pixels. Besides, the proposed method has the ability to well handle the image whose histogram has very unequal peaks and broad valley. [Copyright &y& Elsevier]

Feature selection plays an important role in pattern classification. In this paper, we present an improved branch and bound algorithm for optimal feature subset selection. This algorithm searches for an optimal solution in a large solution tree in an efficient manner by cutting unnecessary paths which are guaranteed not to contain the optimal solution. Our experimental results demonstrate the effectiveness of the new algorithm. [Copyright &y& Elsevier]

This paper proposes an algorithm for solving subsets of typical (canonical) jigsaw puzzles. This algorithm combines shape and image matching with a cyclic “growth” process that tries to place pieces in their correct positions. First, the jigsaw pieces are extracted from the input image. Then, the corner points of the jigsaw pieces are detected. Next, piece classification and recognition are performed based on jigsaw piece models. Connection relationships between pieces are calculated and finally recovered by using boundary shape matching and image merging. We tested this algorithm by employing real-world images containing dozens of jigsaw pieces. The experiment’s results show this algorithm is efficient and effective. [Copyright &y& Elsevier]

In this paper, the efficiency of branch and bound search algorithms for the computation of K nearest neighbors is studied. The most important aspects that influence the efficiency of the search algorithm are: (1) the decomposition method, (2) the elimination rule, (3) the traversal order and (4) the level of decomposition. First, a theoretical derivation of an efficient decomposition method based on principal component analysis is given. Then, different elimination rules and traversal orders are combined resulting in ten different search algorithms. Since the efficiency is strongly dependent on the level of decomposition, this user specified parameter is optimized first. This optimization is realized by a probabilistic model that expresses the total computation time in function of the node traversal cost and the distance computation cost. All comparisons are based on the total computation time for the optimal level of decomposition. [Copyright &y& Elsevier]

Segmentation of image regions based on their texture is a standard problem in image analysis. Once a set of texture features is selected, several algorithms can be applied to segment the image into regions. This paper presents an extension of the watershed algorithm using a vector gradient and multivariate region merging methods. The algorithm uses a set of texture images, and it only depends on an adjustable parameter. Results are presented on a standard set of synthetic images and on textured medical ones, using different texture parameters and merging criteria. [Copyright &y& Elsevier]

The performance of most practical classifiers improves when correlated or irrelevant features are removed. Machine based classification is thus often preceded by subset selection––a procedure which identifies relevant features of a high dimensional data set. At present, the most widely used subset selection technique is the so-called “wrapper” approach in which a search algorithm is used to identify candidate subsets and the actual classifier is used as a “black box” to evaluate the fitness of the subset. Fitness evaluation of the subset however requires cross-validation or other resampling based procedure for error estimation necessitating the construction of a large number of classifiers for each subset. This significant computational burden makes the wrapper approach impractical when a large number of features are present.In this paper, we present an approach to subset selection based on a novel definition of the classifiability of a given data. The classifiability measure we propose characterizes the relative ease with which some labeled data can be classified. We use this definition of classifiability to systematically add the feature which leads to the most increase in classifiability. The proposed approach does not require the construction of classifiers at each step and therefore does not suffer from as high a computational burden as a wrapper approach. Our results over several different data sets indicate that the results obtained are at least as good as that obtained with the wrapper approach. [Copyright &y& Elsevier]

We focus in this paper on a new model for graph decimation which is based on the stochastic pyramid algorithm. In this study, we present some trends towards a better understanding of the new properties arising from the removal of the sequentiality constraint of this so-called technique. [Copyright &y& Elsevier]

In this paper, a new clustering algorithm for the domain of graphs is introduced. Also a number of cluster validation indices are reviewed. These indices aim at finding the optimal number of clusters automatically. The suitability of the considered indices is experimentally evaluated. [Copyright &y& Elsevier]

This paper presents new algorithms to identify and eliminate mislabelled, noisy and atypical training samples for supervised learning and more specifically, for nearest neighbour classification. The main goal of these approaches is to enhance the classification accuracy by improving the quality of the training data. Several experiments with synthetic and real data sets are carried out in order to illustrate the behaviour of the schemes proposed here and compare their performance with that of other traditional techniques. It is also analysed the ability of these new algorithms to “reduce” the possible overlapping among regions of different classes. [Copyright &y& Elsevier]

In this paper, we present an image-adaptive joint photographers expert group encoding algorithm that incorporates global thresholding based on the statistical information of the discrete cosine transform (DCT) coefficient distributions. These statistics are also used to construct rate/distortion-specific quantization tables with nearly optimal tradeoffs. Results on natural and SPOT images demonstrate that the proposed thresholding technique improves the reconstructed image quality compared to that provided by other DCT image coding schemes without thresholding. [Copyright &y& Elsevier]

In this paper, we propose an original approach for the binarization of gray-level images. A multi-scale algorithm based on a statistical test of homogeneity decides if a region belongs to the background or not. At each iteration the image is smoothed with a nonlinear filter to remove the noise. Stable regions in scale space are used as a model to automatically find a threshold from the intensity histogram. [Copyright &y& Elsevier]

Image segmentation has been, and still is, an important research area in Computer Vision, and hundreds of segmentation algorithms have been proposed in the last 30 years. However, elementary segmentation techniques based on either boundary or region information often fail to produce accurate segmentation results on their own. In the last few years, there has therefore been a trend towards algorithms that take advantage of their complementary nature. This paper reviews various segmentation proposals that integrate edge and region information and highlights different strategies and methods for fusing such information. The key objective is to point out the advantages and disadvantages of the various approaches, as well as to comment upon new and interesting ideas that perhaps have not been properly exploited. [Copyright &y& Elsevier]

In this paper, we show that the randomized sampling and voting process detects optical flow. Using an appropriate number of images from a sequence of images, our method detects subpixel motion in this sequence. We use the accumulator space for the unification of these flow vectors which are computed from different time intervals. Numerical examples for the test image sequences show the performance of our method. [Copyright &y& Elsevier]

This paper presents an algorithm for estimating the local fractal dimension (LFD) of textured images. The algorithm is established by an experimental approach based on the blanket method. The proposed method uses the near optimum number of blankets to obtain the LFD for a small local window. The robustness of the proposed method to consistently estimate the LFD using up to a 3×3 local window is confirmed by experimental evaluations. The LFD maps, created from natural scenes, are utilized in an image segmentation algorithm that demonstrates the capability of rough segmentation of fine-texture regions in natural images. [Copyright &y& Elsevier]

This paper presents an efficient region-based motion segmentation method for segmentation of moving objects in a traffic scene with a focus on a video monitoring system (VMS). The presented method consists of two phases: first, in the motion detection phase, the positions of moving objects in a scene are determined using an adaptive thresholding method. To detect varying regions by moving objects, instead of determining the threshold value manually, we use an adaptive thresholding method to automatically choose the threshold value. Second, in the motion segmentation phase, pixels that have similar intensity and motion information are segmented using a weighted k-means clustering algorithm to the binary region of the motion mask obtained in the motion detection. In this way, we need not process a whole image so computation time is reduced. Experimental results demonstrate robustness not only in the variation of luminance conditions and changes in environmental conditions, but also for occlusions among multiple moving objects. [Copyright &y& Elsevier]

Medial axis transform (MAT) is well known for object representation. It is interesting to explore its use in different kinds of computations. In this paper an algorithm has been proposed for computation of normals at the boundaries of two-dimensional objects based on their MATs. In this technique, there is no requirement of linking boundary points during the computation compared to other existing techniques. The added advantage in the computation is that the computation can be restricted purely in the integer domain. [Copyright &y& Elsevier]

Local symmetry is often visible at the corner points. In this paper, generalized symmetry transform (GST) [Reisfeld et al., Internat. J. Comput. Vision 14 (1995) 119] is combined with the proposed parametric corner equation in order to detect corners and improve their localization. Experimental results show that the proposed corner detection algorithm is efficient and robust to noise when tested on the real and artificial images and compared to the other detectors. [Copyright &y& Elsevier]

Most conventional derivative-based corner detectors have shortcomings such as missing junctions, poor localization, sensitivity to noise and high computational cost. This paper presents a new, simple and effective low level processing method to detect corners. In this corner detection algorithm, two oriented cross operators called crosses as oriented pair (COP) are used, which provide useful information to extract low-level features due to its characteristics, preference for edge with different direction and simple direction determination. Fast, accurate and noise-robust corner detection is accomplished with the COP. [Copyright &y& Elsevier]

In this paper, general guidelines and specific algorithms for code transformations between breadth-first (BF) and depth-first (DF) linear quadtrees are proposed. Each algorithm has time and space complexities in O(l), where l is the length of the code of a BF or DF linear quadtree. [Copyright &y& Elsevier]

The original k-means clustering algorithm is designed to work primarily on numeric data sets. This prohibits the algorithm from being directly applied to categorical data clustering in many data mining applications. The k-modes algorithm [Z. Huang, Clustering large data sets with mixed numeric and categorical value, in: Proceedings of the First Pacific Asia Knowledge Discovery and Data Mining Conference. World Scientific, Singapore, 1997, pp. 21–34] extended the k-means paradigm to cluster categorical data by using a frequency-based method to update the cluster modes versus the k-means fashion of minimizing a numerically valued cost. However, as is the case with most data clustering algorithms, the algorithm requires a pre-setting or random selection of initial points (modes) of the clusters. The differences on the initial points often lead to considerable distinct cluster results. In this paper we present an experimental study on applying Bradley and Fayyad''s iterative initial-point refinement algorithm to the k-modes clustering to improve the accurate and repetitiveness of the clustering results [cf. P. Bradley, U. Fayyad, Refining initial points for k-mean clustering, in: Proceedings of the 15th International Conference on Machine Learning, Morgan Kaufmann, Los Altos, CA, 1998]. Experiments show that the k-modes clustering algorithm using refined initial points leads to higher precision results much more reliably than the random selection method without refinement, thus making the refinement process applicable to many data mining applications with categorical data. [Copyright &y& Elsevier]

This paper proposes a new video sequence segmentation method based on the genetic algorithm (GA) that can improve computational efficiency. The computation is distributed into chromosomes that evolve using distributed genetic algorithms (DGAs). Each chromosome consists of a label and feature vector. The label is used as the region number for the pixel where the chromosome is located. Based on the temporal correlation between two consecutive frames in a video sequence, the segmentation of a frame is successively obtained using the segmentation result of the previous frame. In addition to eliminating redundant computation, only unstable chromosomes corresponding to moving object parts are evolved. Experimental results confirm the effectiveness of the proposed method. [Copyright &y& Elsevier]