Mar Pujol

A new formulation of Adomian method

A new approach of the decomposition method (Adomian) in which the Adomian scheme is obtained in a more natural way than in the classical presentation, is given. A new condition for obtaining convergence of the decomposition series is also included.

Face Detection Based on Skin Color Segmentation Using Fuzzy Entropy

Face detection is the first step of any automated face recognition system. One of the most popular approaches to detect faces in color images is using a skin color segmentation scheme, which in many cases needs a proper representation of color spaces to interpret image information. In this paper, we propose a fuzzy system for detecting skin in color images, so that each color tone is assumed to be a fuzzy set. The Red, Green, and Blue (RGB), the Hue, Saturation and Value (HSV), and the YCbCr (where Y is the luminance and Cb,Cr are the chroma components) color systems are used for the development of our fuzzy design. Thus, a fuzzy three-partition entropy approach is used to calculate all of the parameters needed for the fuzzy systems, and then, a face detection method is also developed to validate the segmentation results. The results of the experiments show a correct skin detection rate between 94% and 96% for our fuzzy segmentation methods, with a false positive rate of about 0.5% in all cases. Furthermore, the average correct face detection rate is above 93%, and even when working with heterogeneous backgrounds and different light conditions, it achieves almost 88% correct detections. Thus, our method leads to accurate face detection results with low false positive and false negative rates.

Decomposition method applied to hydrology

In some papers G. Adomian has presented a decomposition technique in order to solve different non‐linear equations. The solution is found as an infinite series quickly converging to accurate solutions. The method is well‐suited for physical problems and it avoids linearization, perturbation and other restrictions, methods and assumptions which may change the problem being solved – sometimes seriously – unnecessarily. Proofs of convergence are given by Cherruault and co‐authors. Many numerical studies for physical phenomena, such as Fisher’s equation, Lorentz’s equation and Edem’s equation are given and solved. In this work, the general equation given by ∂ p \over ∂ t = (∇ ⋅(q(x)⋅ ∇p)) + f(x, t) is solved by using decomposition methods, and is compared to other techniques. This equation can be used to describe the motion of a fluid flow in the so‐called reservoir region, where p(x, t) represents the pressure distribution, f(x, t) describes the withdrawal or injection of the fluid, and q(x) is the transmissibility in the reservoir region.

L-system-driven self-assembly for swarm robotics

In this paper, an assembly swarm algorithm, that will generate microscopic rules from a macroscopic description of complex structures, will be presented. The global structure will be described in a formal way using L-systems (Lindenmayer systems). The proposed algorithm is mainly parallel and exhibit parsimony at microscopic level, being robust and adaptable. In addition, a comparation between a swarm with centralized control and our distributed swarm algorithm will be provided, comparing the time need by the swarm to be assembled and the number of messages exchanged between agents.

Using Gaussian Processes in Bayesian Robot Programming

In this paper, we present an adaptation of Gaussian Processes for learning a joint probabilistic distribution using Bayesian Programming. More specifically, a robot navigation problem will be showed as a case of study. In addition, Gaussian Processes will be compared with one of the most popular techniques for machine learning: Neural Networks. Finally, we will discuss about the accuracy of these methods and will conclude proposing some future lines for this research.

Minimization of an energy function with robust features for image segmentation

In this work, we propose an approach to the model based on Markov random field (MRF) as a systematic way for integrating constraints for robust image segmentation. To do that, robust features and their integration in the energy function, which directs the process, have been defined. The suitability of the method has been verified by comparing classic features with the robust ones. In this approach, the image is first segmented into a set of disjoint regions and the adjacent graph (AG) has been determined. This approach is applied by defining an MRF model on the corresponding AG. Robust features are incorporated to the energy function by means of clique functions, and optimal segmentation is then achieved by finding a labelling configuration, which minimizes the energy function using the simulated annealing.

On searching for an optimal threshold for morphological image segmentation

Segmentation of images represents the first step in many of the tasks that pattern recognition or computer vision has to deal with. Therefore, the main goal of our paper is to describe a new method for image segmentation, taking into account some Mathematical Morphology operations and an adaptively updated threshold, what we call Morphological Gradient Threshold, to obtain the optimal segmentation. The key factor in our work is the calculation of the distance between the segmented image and the ideal segmentation. Experimental results show that the optimal threshold is obtained when the Morphological Gradient Threshold is around the 70% of the maximum value of the gradient. This threshold could be computed, for any new image captured by the vision system, using a properly designed binary metrics.

Applying distance histograms for robust object recognition

Purpose – One of the main goals of vision systems is to recognize objects in real world to perform appropriate actions. This implies the ability of handling objects and, moreover, to know the relations between these objects and their environment in what we call scenes. Most of the time, navigation in unknown environments is difficult due to a lack of easily identifiable landmarks. Hence, in this work, some geometric features to identify objects are considered. Firstly, a Markov random field segmentation approach is implemented. Then, the key factor for the recognition is the calculation of the so‐called distance histograms, which relate the distances between the border points to the mass center for each object in a scene.Design/methodology/approach – This work, first discusses the features to be analyzed in order to create a reliable database for a proper recognition of the objects in a scene. Then, a robust classification system is designed and finally some experiments are completed to show that the reco...

Agent-based recognition of facial expressions

Description of a system to detect facial expressions using an agent-based approach is presented. The system utilizes interaction between Matlab-based image filters and a JADE-based agent implementation. The system is demonstrated using a feature recognition example. The system however has a much wider applicability, especially as Matlab is used extensively in other scientific/numerical computing applications.

A cybernetic approach to the multiscale minimization of energy function: Grey level image segmentation

Segmentation is an important topic in computer vision and image processing. In this paper, we sketch a scheme for a multiscale segmentation algorithm and prove its validity on some real images. We propose an approach to the model based on MRF (Markov Random Field) as a systematic way for integrating constraints for robust image segmentation. To do that, robust features and their integration in the energy function, which directs the process, have been defined. In this approach, the image is first transformed to different scales to determine which one fits better to our purposes. Then, it is segmented into a set of disjoint regions, the adjacent graph (AG) is determined and a MRF model is defined on the corresponding AG. Robust features are incorporated to the energy function by means of clique functions and optimal segmentation is then achieved by finding a labeling configuration that minimizes the energy function using Simulated Annealing.