Catálogo de publicaciones - libros

The paper proposes a solution for the optimization of traveling times in a metropolitan area that exploits the traffic images collected from webcams located at the crossroads of the traffic network. This is achieved by optimizing the traffic light cycles according to a distributed mathematical model whose solution is obtained by using suitable soft-computing agents resident on the nodes of the information network where they are responsible of processing the webcam images and of managing the traffic light cycles.

An efficient and straightforward method to compute the variance in the Maximum Variance Cluster algorithm is proposed. Experiments on both artificial and real data sets have been performed. Both the time of execution and the quality of the results were considered. According to the results, the method proposed speeds the MVC algorithm up to 50‰ without loosing its accurancy on the majority of the data sets considered.

The following paper presents results of research on the application of deformable grids in image analysis for the purpose of speech reading. The main contribution of the paper is a modification of the original paradigm through an extension of a pool of parameters used for evaluation of gridmatching result. This significantly improves classification performance and indicates that the adopted approach to image analysis offers reasonable representation of information on uttered phones.

In this paper the Pattern Recognition System is described that is designed for features extraction, classifiers learning and classification. Patterns may be presented either in graphical or numerical form. The system is created in a user-friendly manner and enables to carry on some experiments with different learning algorithms. Its additional advantages are enhanced capabilities of graphical presentation and visualization of the learning process and it's results. The system is an universal tool that can be applied in didactics, lectures and in some research work as well.

We describe an algorithm for matching two-dimensional coordinate lists. Such matching should be immune to translation, rotation, scaling and flipping. We assume that coordinate lists may be only partially overlapping and we allow some random perturbations in the lists. Our goal is to match enough points to be able to derive a coordinates transformation with high confidence. The implementation described here is a part of SkySpy automated sky survey system. The presented algorithm is used to perform matching of stars detected during observations with star catalogues.

In this paper, a new statistical projection-based method called Two-Dimensional- Oriented Linear Discriminant Analysis (2DO-LDA) is presented. While in the Fisherfaces method the 2D image matrices are first transformed into 1D vectors by merging their rows of pixels, 2DO-LDA is directly applied on matrices, as 2D-PCA. Within and between-class image covariance matrices are generalized, and 2DO-LDA aims at finding a projection space jointly maximizing the second and minimizing the first by considering a generalized Fisher criterion defined on image matrices. A series of experiments was performed on various face image databases in order to evaluate and compare the effectiveness and robustness of 2DO-LDA to 2D-PCA and the Fisherfaces method. The experimental results indicate that 2DO-LDA is more efficient than both 2D-PCA and LDA when dealing with variations in lighting conditions, facial expression and head pose.

This paper presents a novel method to reconstruct 3D body surface model of human figure from 2D uncalibrated monocular video. Unlike other works on body reconstruction from monocular video where either the source video are strictly selected or the skeleton proportions are assumed available beforehand, the proposed system automatically recovers the skeleton proportions and rebuild the 3D human body shape in NURBS format without pose estimation.

Mura – irregular lightness variation on uniformly manu-factured surface – is required to be detected to keep high quality of the display devices. The mura is perceived in spite of its low contrast, if its spatial frequency falls on a sensitive range of human vision. We propose a method considering characteristics of human vision to detect mura of the display devices’ components that have lower intensity than the final device.

This paper presents a method for interactively animating characters that travel over uneven virtual terrain. The method is based on algorithmically generating limb and pelvis trajectories and animating the skeleton with a custom Inverse Kinematics algorithm. Simple but naturally looking and highly customizable models for feet and pelvis trajectories are proposed. Fluent blending of different motions with the proposed approach is possible via linear parameter blending. The algorithm is fast, has very low memory requirements and needs no expensive preprocessing.

We introduce medium precision avatars in VRML worlds for cultural heritage presented on-line. Avatars are integration of UI or they appear in the cyber city on the places where interesting stories can be told to the user. Avatars in medium distance communication use empathy for better delivering of message. We propose a new “delayed mirror” metaphor for empathic communication in virtual reality for cultural heritage. It combines level of detail, real-time interaction and digital storytelling. Avatar’s face structure is based on minimal Perlin’s face. This precision is intended for outdoor scenes and medium distance communication. However, the facial expressions of a virtual tourist can be captured by a cheap hardware set-up consisting from a pair of web cams, and employed for improving the empathy. For body motions and deformations in VRML we use H-Anim standard5.