Search

Abstract: In some recent papers we have introduced a notion of continuity in digital spaces which extends the usual notion of digital continuity. Our approach, which uses multivalued maps, provides a better framework to define topological notions, like retractions, in a far more realistic way than by using just single-valued digitally continuous functions. In particular, we have characterized the deletion of simple points and some well known parallel thinning algorithm as a particular type of retractions, called -retractions. In this paper we give a full characterization of the thinning algorithms that can be modeled as -retractions. This class agrees, with minor modifications, with the class of thinning algorithms satisfying Ronse sufficient conditions for preservation of topology. [Copyright &y& Elsevier]

The appearance model is an important issue in the visual tracking. In this paper, we present a structured partial least squares (SPLS) based appearance model for object tracking which not only discriminates the target from its surrounding background but also able to tolerate its appearance variations. The target is represented as a low-dimensional feature vector, which is the structured combination of the subspaces learned by SPLS. In order to capture the variations in appearance during the tracking, we update the appearance model by taking the estimated class labels of the new targets obtained online into account. Furthermore, aiming to alleviate the drift problems, negative samples generated by disordering the structured appearance model are added in the training samples. Both qualitative and quantitative evaluations on challenging benchmark video sequences demonstrate that the proposed tracking algorithm performs favorably against several state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Abstract: In this paper scheduling communications in data gathering networks is analyzed. We study collecting information by a set of sensors, each of which stores the data in its memory buffer and then passes them to a base station. The network lifetime ends as soon as the first node is out of memory. We use a divisible load model to propose a communication scheduling algorithm that maximizes the system lifetime, and hence, the total amount of gathered information. The influence of the sensor communication rates on the obtained schedules is exposed in a series of computational experiments. [Copyright &y& Elsevier]

Abstract: Construction of Space Filling Designs in high dimensional space remains difficult since powerful algorithms at low dimensions become difficult to use at higher dimensions that leads to non-uniform distribution in the factor space. We propose in this paper two approaches in order to repair designs: Curvilinear Component Analysis (CCA) and the Wootton, Sergent, Phan-Tan-Luu's algorithm called WSP in order to detect clusters and to fill gaps. Thus, CCA allows visualization of two or more very closely-spaced points in D dimensions by projecting them in a 2 dimensions space. Then identified clusters can be eliminated using the WSP algorithm. Moreover, the presence of gaps in input space could be very problematic since no information on the phenomenon is available and the WSP algorithm will be used in order to fill gaps by adding points in the “empty” zones. A new quality criterion has been proposed in order to follow the reparation steps. Examples in different dimensions are presented to illustrate these methods. [Copyright &y& Elsevier]

Abstract: Model identification plays a central role in any activity associated with process operations. With control being done on different levels, different models are required for the same plant, each for a different range of dynamics. Besides that most identification methods apply to the linear models, they also do not allow for selecting a frequency range. Wavelet-based methods have the intrinsic ability to select time and frequency windows, and, to some extent, are also applicable to non-linear processes. The paper presents an approach, in which the wavelet transform is employed for system identification enabling the selection of the particular frequency range of interest. We will show the use of some wavelet filters with a property of superior selectivity in the frequency domain and having compact support in the time domain, which, in turn, influences an accurate implementation. These properties provide us with a possibility of the measured data analysis in the frequency domain without any loss of information. Selection of a proper filter allows us to identify the system on a desired frequency range, or to identify a number of systems for distinct frequency ranges. This is specifically convenient for the systems with dominant modes, such as singularly perturbed systems. The possibility of selection of the specific frequency range can be utilized for application-based identification such as control, when only a limited frequency range is required. We conclude with a case study, where the proposed algorithms are tested and results are presented. [Copyright &y& Elsevier]

Abstract: In this paper, we propose the MIML (Multi-Instance Multi-Label learning) framework where an example is described by multiple instances and associated with multiple class labels. Compared to traditional learning frameworks, the MIML framework is more convenient and natural for representing complicated objects which have multiple semantic meanings. To learn from MIML examples, we propose the MimlBoost and MimlSvm algorithms based on a simple degeneration strategy, and experiments show that solving problems involving complicated objects with multiple semantic meanings in the MIML framework can lead to good performance. Considering that the degeneration process may lose information, we propose the D-MimlSvm algorithm which tackles MIML problems directly in a regularization framework. Moreover, we show that even when we do not have access to the real objects and thus cannot capture more information from real objects by using the MIML representation, MIML is still useful. We propose the InsDif and SubCod algorithms. InsDif works by transforming single-instances into the MIML representation for learning, while SubCod works by transforming single-label examples into the MIML representation for learning. Experiments show that in some tasks they are able to achieve better performance than learning the single-instances or single-label examples directly. [Copyright &y& Elsevier]

Abstract: The original local binary fitting (LBF) model is sensitive to contour initialization and thus easily obtains an inaccurate result due to improper initialization. This paper presents a new method that not only can arrive at sub-pixel accuracy, but also allows for more flexible initialization of the contour. Two important terms play main role in our new method. One is an image gradient alignment term (IGA) which uses the directional information of the image gradient, the other is a local intensity fitting term (LIF) which makes use of local region information. The integration of the above two terms prevents our method from being sensitive to contour initialization. In addition, a global intensity fitting term (GIF) multiplied by a stopping function is included, which can speed up our algorithm while do not influence the accuracy of the segmentation result. Using the simple central difference, the gradient descend flow equation for the level set function can be easily and efficiently implemented. The results on several synthetic and real images demonstrate the effectiveness and accuracy of our method. [Copyright &y& Elsevier]

Abstract: The traditional particle filter algorithm cannot solve the validity of particles ideally. In this situation, a large amount of particles are required to guarantee tracking performance and tracking windows cannot change scale with targets adaptively. In addition, the particle filter algorithm which uses a single cue cannot achieve stable tracking regarding sudden illumination variation and similarly coloured background clutters. In this paper, an algorithm that combines CamShift with particle filter using multiple cues is proposed. The effectiveness of particles is improved and the tracking window can change scale with the target adaptively because of the use of CamShift. At the same time, an adaptive integration method is used to combine colour information with motion information, so the problems can be solved which are encountered in tracking an object with illumination variation and the background color clutter. Moreover, an occlusion handler is proposed to handle the full occlusion for a long time. [ABSTRACT FROM AUTHOR]

Abstract: This paper deals with the stabilizability of interconnected systems via linear time-invariant (LTI) decentralized controllers. Given a controllable and observable system with some distinct decentralized fixed modes (DFMs), it is desired to find a suitable control structure (in terms of information flow) for it. Since a decentralized controller consists of a number of non-interacting local controllers, the objective here is to establish certain interactions between the local controllers in order to eliminate the undesirable DFMs. This objective is achieved by translating the knowledge of the system into some bipartite graphs. Then, the notions of minimal sets and maximal subgraphs are introduced, which lead to a simple combinatorial algorithm for solving the underlying problem. The efficacy of the results obtained is demonstrated by an illustrative example. [Copyright &y& Elsevier]