Catálogo de publicaciones - libros

Based on the directional characteristic of SAD(Sum of Absolute Difference) distribution and the center-biased characteristic of motion vectors, a fast BMA(block-matching motion estimation algorithm), DSSS(Diamond and Simplified Square Search), is proposed in this paper. DSSS employs line search pattern(LP), triangle search pattern(TP), or square pattern(SP) adaptively according to the distance between the MBD(Minimum Block Distortion) and SMBD(Second MBD) points to locate the best matching block with large motion vector, and diamond search pattern(DP) to refine the motion vector. Although the proposed DSSS may also be trapped in local minima, the experimental results show that it is faster than DS(Diamond Search) and DTS(Diamond and Triangle Search), while its encoding efficiency is better than DS and it is almost the same as that of DTS.

In this paper we introduce a new method for selecting prototypes with Mixed Incomplete Data (MID) object description, based on an extension of the Nearest Neighbor rule. This new rule allows dealing with functions that are not necessarily dual functions of distances. The introduced compact set editing method (CSE) constructs a prototype consistent subset, which is also subclass consistent. The experimental results show that CSE has a very nice computational behavior and effectiveness, reducing around 50% of prototypes without appreciable degradation on accuracy, in almost all databases with more than 300 objects.

An automatic texture segmentation approach is presented in this paper, in which wavelet-domain hidden Markov tree (WD-HMT) model is exploited to characterize the texture features of an image, an effective cluster validity index, the ratio of the overlap degree to the separation one between different fuzzy clusters, is used to determine the true number of the textures within an image by solving the minimum of this index in terms of different number of clusters, and the possibilistic C-means (PCM) clustering is performed to extract the training sample data from different textures. In this way, unsupervised segmentation is changed into self-supervised one, and the well-known HMTseg algorithm in the WD-HMT framework is eventually used to produce the final segmentation results, consequently automatic segmentation process is completed. This new approach is applied to segment a variety of composite textured images into distinct homogeneous regions with satisfactory segmentation results demonstrated. Real-world images are also segmented to further justify our approach.

The KNN rule has been widely used in many pattern recognition problems, but it is sensible to noisy data within the training set, therefore, several sample edition methods have been developed in order to solve this problem. A. Franco, D. Maltoni and L. Nanni proposed the Reward-Punishment Editing method in 2004 for editing numerical databases, but it has the problem that the selected prototypes could belong neither to the sample nor to the universe. In this work, we propose a modification based on selecting the prototypes from the training set. To do this selection, we propose the use of the Fuzzy C-means algorithm for mixed data and the KNN rule with similarity functions. Tests with different databases were made and the results were compared against the original Reward-Punishment Editing and the whole set (without any edition).

Stable coordinate pairs (SCP) like ‘comments and suggestions’ or ‘safe and sound’ are rather frequent in texts in Spanish, though there are only few thousands of them in language. We characterize SCPs statistically by a numerical Stable Connection Index and reveal its unimodal distribution. We also propose lexical, morphologic, syntactic, and semantic categories for SCP structural description — for both a whole SCP and its components. It is argued that database containing a set of categorized SCPs facilitates several tasks of automatic NLP.. The research is based on a set of ca. 2200 Spanish coordinate pairs.

was used as model to establish an index for oocyte maturity determination in zooplankters based in citometry and histochemical evaluation of gonadic masses. Biometry was performed using an ocular micrometer and nucleus/cytoplasm ratios were obtained characterizing each of the three identified stages: Immature, Vitellogenic and Mature. This paper presents a novel approach since it joins (and, indeed, reinforces) the index framework with the evaluation of the same biological samples by a suitable combination of deformable models. Nucleus contour is identified through Active Shape Models techniques, and cytoplasm contour’s detected through parametric Snakes, with prior image preprocessing based on statistical and mathematical morphology techniques. Morphometric parameters such as nucleus and cytoplasm area and ratio between them are then easily computed. As a result the dataset validated the applied methodology with a realistic background and a new, more accurate and ecologically realistic index for oocyte staging emerged.

Most traumatology services use radiological images to control the state and possible displacements of total hip replacement implants. Prostheses are typically and traditionally detected by means of edge detectors, a widely used technique in medical image analysis. This article analyses how different edge detectors identify the prosthesis in X-Rays by measuring the performance of each detection algorithm; it also determines the clinical usefulness of the algorithms with the help of clinical experts.

In this paper, we present an incremental clustering algorithm in the logical combinatorial approach to pattern recognition, which finds incrementally the -compact sets with radius of an object collection. The proposed algorithm allows generating an intermediate subset of clusters between the -connected components and -compact sets (including both of them as particular cases). The evaluation experiments on standard document collections show that the proposed algorithm outperforms the algorithms that obtain the -connected components and the -compact sets.

Image registration is a fundamental task in image processing. Its aim is to match two or more pictures taken with the same or from different sensors, at different times or from different viewpoints. In image registration the use of an adequate measure of alignment is a crucial issue. Current techniques are classified in two broad categories: pixel based and feature based. All methods include some similarity measure. In this paper a new measure that combines mutual information ideas, spatial information and feature characteristics, is proposed. Edge points obtained from a Canny edge detector are used as features. Feature characteristics like location, edge strength and orientation, are taken into account to compute a joint probability distribution of corresponding edge points in two images. Mutual information based on this function is maximized to find the best alignment parameters. The approach has been tested with a collection of medical images (Nuclear Magnetic Resonance and radiotherapy portal images) and conventional video sequences, obtaining encouraging results.

In this paper, an adaptive scene-based nonuniformity correction methodology for infrared image sequences is developed. The method estimates detector parameters and carry out the non-uniformity correction based on the recursive least square filter approach, with adaptive supervision. The key advantage of the method is based in its capacity for estimate detectors parameters, and then compensate for fixed-pattern noise in a frame by frame basics. The ability of the method to compensate for nonuniformity is demonstrated by employing several infrared video sequences obtained using two infrared cameras.