Pattern recognition methods are used in the final stage of a traffic sign detection and recognition system, where the main objective is to categorize a detected sign. Support vector machines have been reported as a good method to achieve this main target due to their ability to provide good accuracy as well as being sparse methods. Nevertheless, for complete data sets of traffic signs the number of operations needed in the test phase is still large, whereas the accuracy needs to be improved. The objectives of this work are to propose pre-processing methods and improvements in support vector machines to increase the accuracy achieved while the number of support vectors, and thus the number of operations needed in the test phase, is reduced. Results show that with the proposed methods the accuracy is increased 3-5% with a reduction in the number of support vectors of 50-70%.

No abstract

No abstract

No abstract

Abstract-From this paper, we propose a novel methodology to compute a 2D Homography applying some algorithms of computer algebra. We consider the classical problem of solving (exactly) a linear system of algebraic equations, and we suggest a new algorithm for computer vision, based on homomorphism methods over Z, to solve a system of equations necessary to achieve a 3 × 3 matrix H which lets us to compute the projective transformation which translates coordinates between points in different planes. From this work, we want to show that it is possible to apply a symbolic approach to some crucial issues of computer vision, moreover of the numerical methodology, in order to reduce the complexity of some algorithms, and to eliminate the problems associated with loss of precision and normalization. We test our technique in a real situation: a parking management system, which creates a pseudo-top-view of a parking area to determine if there are free parking lots or not.

Cyclic voltammetry is an electroanalytical technique for obtaining information about substances under analysis without the need for complex flow systems. However, classifying the information in voltammograms obtained using this technique is difficult. In this paper, we propose the use of fixed kernel regression as a method for extracting features from these voltammograms, reducing the information to a few coefficients. The proposed approach has been applied to a wine classification problem with accuracy rates of over 98%. Although the method is described here for extracting voltammogram information, it can be used for other types of signals.

Abstract. Recent works in object recognition often use visual words, i.e. vector quantized local descriptors extracted from the images. In this paper we present a novel method to build such a codebook with class representative vectors. This method, coined Cluster Precision Maximization (CPM), is based on a new measure of the cluster precision and on an optimization procedure that leads any clustering algorithm towards class representative visual words. We compare our procedure with other measures of cluster precision and present the integration of a Reciprocal Nearest Neighbor (RNN) clustering algorithm in the CPM method. In the experiments, on a subset of the the Caltech101 database, we analyze several vocabularies obtained with different local descriptors and different clustering algorithms, and we show that the vocabularies obtained with the CPM process perform best in a category-level object recognition system using a Support Vector Machine (SVM).

No abstract