Catálogo de publicaciones - libros

In this paper we propose biometric descriptors inspired by shape signatures traditionally used in graphics recognition approaches. In particular several methods based on line shape descriptors used to identify newborns from the biometric information of the ears are developed. The process steps are the following: image acquisition, ear segmentation, ear normalization, feature extraction and identification. Several shape signatures are defined from contour images. These are formulated in terms of zoning and contour crossings descriptors. Experimental results are presented to demonstrate the effectiveness of the used techniques.

The Arc Segmentation Contest, as the sixth in the series of graphics recognition contests organized by IAPR TC10, was held in association with the GREC’2005 workshop. Three systems have participated in the contest. In this paper we present a brief summary: the contest rules, the updated performance metrics, test images and their ground truths, and the outcomes.

This paper provides some information regarding the winning system at the GREC’2005 contest on arc segmentation. Important facts are first recalled, then the changes made on the system since its first presentation at GREC’2001 are detailed. The obtained results are briefly commented, and the paper finally concludes with some clues for possible, future improvements regarding the system.

The robust detection of lines and arcs in scanned documents or technical drawings is an important problem in document image understanding. We present a new solution to this problem that works directly on run-length encoded data. The method finds globally optimal solutions to parameterized thick line and arc models. Line thickness is part of the model and directly used during the matching process. Unlike previous approaches, it does not require any thinning or other preprocessing steps, no computation of the line adjacency graphs, and no heuristics. Furthermore, the only search-related parameter that needs to be specified is the desired numerical accuracy of the solution. The method is based on a branch-and-bound approach for the globally optimal detection of these geometric primitives using runs of black pixels in a bi-level image. We present qualitative and quantitative results of the algorithm on images used in the 2003 and 2005 GREC arc segmentation contests.

Following the experience of the first edition of the international symbol recognition contest held during GREC’03 in Barcelona, a second edition has been organized during GREC’05. In this paper, first, we bring to mind the general principles of both contests before presenting more specifically the details of this last edition. In particular, we describe the dataset used in the contest, the methods that took part in it, and the analysis of the results obtained by the participants. We conclude with a synthesis of the contributions and lacks of these two editions, and some leads for the organization of a forthcoming contest.

In this paper, we present an integrated system for symbol recognition. The whole recognition procedure consists of image compression, denoising and recognition. We present a pixel-based method to calculate similarity between two symbols using the after they are aligned by their angular distributions. The proposed method can overcome some shortcomings of other pixel-level methods. We also propose a new denoising technique in our system to improve the recognition precision and efficiency. Evaluation results on test sets provided by the 2nd IAPR contest on symbol recognition show good performance of the system in recognizing symbols with degradation and affine transformation.

In this paper, we propose a robust moment invariant which has a higher discriminant factor based on Fisher linear discriminant analysis that can deal with noise degradation, deformation of vector distortion, translation, rotation and scale invariant. The proposed system for the symbol recognition consists of 3 steps: 1) degradation model preprocessing step, 2) a different normalization for the second moment invariant and a measure for roundness and eccentricity for feature extraction step, 3) k-Nearest Neighbor with Mahalanobis distance compared to Euclidean distance and k-D tree for classifier. A comparison using multi-layer feed forward neural network classifier is given. An improvement of the discriminant factor around 4 times is achieved compared to that of the original normalized second moments using GREC 2005 dataset. Experimentally we tested our system with 3300 training images using k-NN classifier and on all 9450 images given in the dataset and achieved recognition rates higher than 86 % for all degradation models and 96 % for degradation models 1 to 4.

GREC’05 marked the 10th anniversary of the series of international workshops on graphics recognition, for which the first edition was held in Penn State in 1995. At the end of the workshop, a panel discussion was held to take a broader view of our field, to examine the shift in issues addressed by research teams, and define some challenges for the coming years. This paper tries to summarize the results of the panel discussion.