Enrique Domínguez

FOREGROUND DETECTION IN VIDEO SEQUENCES WITH PROBABILISTIC SELF-ORGANIZING MAPS

Background modeling and foreground detection are key parts of any computer vision system. These problems have been addressed in literature with several probabilistic approaches based on mixture models. Here we propose a new kind of probabilistic background models which is based on probabilistic self-organising maps. This way, the background pixels are modeled with more flexibility. On the other hand, a statistical correlation measure is used to test the similarity among nearby pixels, so as to enhance the detection performance by providing a feedback to the process. Several well known benchmark videos have been used to assess the relative performance of our proposal with respect to traditional neural and non neural based methods, with favourable results, both qualitatively and quantitatively. A statistical analysis of the differences among methods demonstrates that our method is significantly better than its competitors. This way, a strong alternative to classical methods is presented.

A neural model for the p-median problem

There exist several neural techniques for solving NP-hard combinatorial optimization problems. At the beginning of the 1980s, recurrent neural networks were shown to be able to solve optimization problems. Criticism of this approach includes the tendency of recurrent neural networks to produce infeasible solutions and poor local minima. This paper proposes a new technique which always provides feasible solutions and removes the tuning phase since the constraints are incorporated in the neural architecture instead of the energy function, therefore the tuning parameters are unnecessary. One of the most popular and well-known facility location problems is the p-median problem, which concerns the location of p facilities (or medians) in order to minimize the total weighted distance between the demand points and the facilities. There exist several heuristics to find optimal solutions of the problem based on the traditional formulation. In this paper a new formulation for the p-median problem based on two types of decision variables and with only n+p linear equality constraints is presented, where n is the number of demand points or customers and p is the number of facilities (medians). Also, a competitive recurrent neural network is proposed for this problem. The neural network consists of two layers (allocation layer and location layer) with 2np process units. The process units constitute n+p groups, where only one process unit per group is active at the same time and process units of the same layer are updated in parallel. Moreover, the energy function (objective function) always decreases or remains constant as the system evolves according to the dynamical rule proposed. The effectiveness and efficiency of our algorithm for different problem sizes are analyzed in comparison to conventional heuristic methods. The results show that our recurrent neural network generates good solutions with a reasonable computational effort. Scope and purpose: Geographical information systems (GIS) have occupied the attention of many researches involving a number of academic fields including geography, civil engineering, computer science, land use planning, and environmental sciences. GIS can support a wide range of spatial queries that can be used to support location studies. Model application and model development are the major impact of GIS on the field of location science. These systems are designed to store, retrieve, manipulate, analyze, and map geographical data. GIS can serve as the source of input data for a location model and it can also be used to present the model results. For example, if a p-median problem has to be solved, then a GIS executes a heuristic algorithm that reads the data from the GIS and it presents the results in real time. Location-allocation models simultaneously locate facilities and allocate demand points to them. These models also arise in a variety of public and private sector problems. The p-median is the most widely used location-allocation model. The p-median model is NP-hard and its data set became very large in real problem, so heuristic solutions are required. As the size of the data set grow, the number of feasible solutions grows and the quality of solutions and computation times from the most commonly used heuristic are deteriorated. The purpose of this paper is to develop a neural model to be integrated in GIS software. Thus, we propose a new formulation for the p-median problem based on 2np binary variables and n+p equality linear constraints. A recurrent neural model is proposed for solving the p-median problem based on this formulation without the difficulty in selecting appropriate tuning parameters, since these parameters are avoided. Moreover, an np-parallel algorithm has been developed based on this formulation. The central property of this algorithm is that the objective function always decreases (or remains constant) as the algorithm evolves according to its dynamical rule. Moreover, this algorithm can be also implemented using optical or VLSI technology.

A Neural Network Approach for Video Object Segmentation in Traffic Surveillance

This paper presents a neural background modeling based on subtraction approach for video object segmentation. A competitive neural network is proposed to form a background model for traffic surveillance. The unsupervised neural classifier handles the segmentation in natural traffic sequences with changes in illumination. The segmentation performance of the proposed neural network is qualitatively examined and compared to mixture of Gaussian models. The proposed algorithm is designed to enable efficient hardware implementation and to achieve real-time processing at great frame rates.

Hierarchical Color Quantization Based on Self-organization

In this paper, a new hierarchical color quantization method based on self-organizing maps that provides different levels of quantization is presented. Color quantization (CQ) is a typical image processing task, which consists of selecting a small number of code vectors from a set of available colors to represent a high color resolution image with minimum perceptual distortion. Several techniques have been proposed for CQ based on splitting algorithms or cluster analysis. Artificial neural networks and, more concretely, self-organizing models have been usually utilized for this purpose. The self-organizing map (SOM) is one of the most useful algorithms for color image quantization. However, it has some difficulties related to its fixed network architecture and the lack of representation of hierarchical relationships among data. The growing hierarchical SOM (GHSOM) tries to face these problems derived from the SOM model. The architecture of the GHSOM is established during the unsupervised learning process according to the input data. Furthermore, the proposed color quantizer allows the evaluation of different color quantization rates under different codebook sizes, according to the number of levels of the generated neural hierarchy. The experimental results show the good performance of this approach compared to other quantizers based on self-organization.

Comparative analysis of modern optimization tools for the p-median problem

This paper develops a study on different modern optimization techniques to solve the p-median problem. We analyze the behavior of a class of evolutionary algorithm (EA) known as cellular EA (cEA), and compare it against a tailored neural network model and against a canonical genetic algorithm for optimization of the p-median problem. We also compare against existing approaches including variable neighborhood search and parallel scatter search, and show their relative performances on a large set of problem instances. Our conclusions state the advantages of using a cEA: wide applicability, low implementation effort and high accuracy. In addition, the neural network model shows up as being the more accurate tool at the price of a narrow applicability and larger customization effort.

A New GHSOM Model Applied to Network Security

The self-organizing map (SOM) have shown to be successful for the analysis of high-dimensional input data as in data mining applications such as network security. However, the static architecture and the lack of representation of hierarchical relations are its main drawbacks. The growing hierarchical SOM (GHSOM) address these limitations of the SOM. The GHSOM is an artificial neural network model with hierarchical architecture composed of independent growing SOMs. One limitation of these neural networks is that they just take into account numerical data, even though symbolic data can be present in many real life problems. In this paper a new GHSOM model with a new metric incorporing both numerical and symbolic data is proposed. This new GHSOM model is proposed for detecting network intrusions. An intrusion detection system (IDS) monitors the IP packets flowing over the network to capture intrusions or anomalies. One of the techniques used for anomaly detection is building statical models using metrics derived from observation of the users actions. Randomly selected subsets that contains both attacks and normal records from the KDD Cup 1999 benchmark are used for training the proposed GHSOM. Experimental results are provided and compared to other hierarchical neural networks.

A self-organizing map to improve vehicle detection in flow monitoring systems

The obtaining of perfect foreground segmentation masks still remains as a challenging task in video surveillance systems, since errors in that initial stage could lead to misleadings in subsequent tasks as object tracking and behavior analysis. This work presents a novel methodology based on self-organizing neural networks and Gaussian distributions to detect unusual objects in the scene, and to improve the foreground mask handling occlusions between objects. After testing the proposed approach on several traffic sequences obtained from public repositories, the results demonstrate that this methodology is promising and suitable to correct segmentation errors on crowded scenes with rigid objects.

A Competitive Neural Network for Multiple Object Tracking in Video Sequence Analysis

Tracking of moving objects in real situation is a challenging research issue, due to dynamic changes in objects or background appearance, illumination, shape and occlusions. In this paper, we deal with these difficulties by incorporating an adaptive feature weighting mechanism to the proposed growing competitive neural network for multiple objects tracking. The neural network takes advantage of the most relevant object features (information provided by the proposed adaptive feature weighting mechanism) in order to estimate the trajectories of the moving objects. The feature selection mechanism is based on a genetic algorithm, and the tracking algorithm is based on a growing competitive neural network where each unit is associated to each object in the scene. The proposed methods (object tracking and feature selection mechanism) are applied to detect the trajectories of moving vehicles in roads. Experimental results show the performance of the proposed system compared to the standard Kalman filter.

An Intrusion Detection System Based on Hierarchical Self-Organization

An intrusion detection system (IDS) monitors the IP packets flowing over the network to capture intrusions or anomalies. One of the techniques used for anomaly detection is building statistical models using metrics derived from observation of the user’s actions. A neural network model based on self organization is proposed for detecting intrusions. The self-organizing map (SOM) has shown to be successful for the analysis of high-dimensional input data as in data mining applications such as network security. The proposed growing hierarchical SOM (GHSOM) addresses the limitations of the SOM related to the static architecture of this model. The GHSOM is an artificial neural network model with hierarchical architecture composed of independent growing SOMs. Randomly selected subsets that contain both attacks and normal records from the KDD Cup 1999 benchmark are used for training the proposed GHSOM.

Bregman divergences for growing hierarchical self-organizing networks.

Growing hierarchical self-organizing models are characterized by the flexibility of their structure, which can easily accommodate for complex input datasets. However, most proposals use the Euclidean distance as the only error measure. Here we propose a way to introduce Bregman divergences in these models, which is based on stochastic approximation principles, so that more general distortion measures can be employed. A procedure is derived to compare the performance of networks using different divergences. Moreover, a probabilistic interpretation of the model is provided, which enables its use as a Bayesian classifier. Experimental results are presented for classification and data visualization applications, which show the advantages of these divergences with respect to the classical Euclidean distance.