Catálogo de publicaciones - libros

An algorithm of estimation of the curvature at each point of a general discrete curve in (log ) is proposed. It uses the notion of blurred segment, extending the definition of segment of arithmetic discrete line to be adapted to noisy curves. The proposed algorithm relies on the decomposition of a discrete curve into maximal blurred segments also presented in this paper.

We present an algorithm for the robust fitting of objects given as voxel data with affinely transformed superellipsoids. Superellipsoids cover a broad range of various forms and are widely used in many application fields. Our approach uses the method of normalization and a new separation of the affine transformation into a shearing, an anisotropic scale and a rotation. It extends our previous work for 2D fitting problems and for fitting rectangular boxes in 3D. Our technique can be used as a valuable tool for solving this fitting task for 3D data. If the exponents describing the superellipsoids to be fitted are known in advance, the method is extremely robust even against major distortions of the object to be fitted.

We address the problem of moving object reconstruction. Several methods have been published in the past 20 years including stereo reconstruction as well as multi-view factorization methods. In general, reconstruction algorithms estimate the 3D structure of the object and the camera parameters in a non-optimal way and then a nonlinear optimization method refines the estimated camera parameters and 3D object coordinates.

In this paper, an adjustment method is proposed which is the fast version of the well-known down-hill alternation method. The novelty which yields the high speed of the algorithm is that the steps of the alternation give optimal solution to the subproblems by closed-form formulas. The proposed algorithm is discussed here and it is compared to the widely used bundle adjustment method.

Given a cell complex whose geometric realization || is embedded in and a continuous function : ||→ (called the ), we construct a graph () which is an extension of the Reeb graph (||). More concretely, the graph () without loops is a subdivision of (||). The most important difference between the graphs () and (||) is that () preserves not only the number of connected components but also the number of “tunnels” (the homology generators of dimension 1) of . The latter is not true in general for (||). Moreover, we construct a map : ()→ identifying representative cycles of the tunnels in with the ones in () in the way that if is a loop in (), then () is a cycle in such that all the points in |()| belong to the same level set in ||.

For quality inspection of printing systems it is necessary to measure the displacement between printing processes. Tie points are employed in correspondence and displacement estimation between individual print elements. We compare interest point and region descriptors for tie point detection in industrial inspection tasks. Clustering of measured displacements taken from sequences of sample images allows the estimation of the accuracy of printing processes and the alignment of printing processes. Results of an experimental application to banknote printing process inspection are given.

In this paper, a method of semi-automatic training set acquisition for character classifiers used in cursive handwriting recognition is described. The training set consists of character samples extracted from a training corpus by segmentation. The method first splits the word images from the corpus into a sequence of graphemes. Then, the set of candidate segmentation variants is elicited with an evolutionary algorithm, where the segmentation variant determines subdivision of grapheme sequences of words into subsequences corresponding to consecutive letters. Segmentation variants are modeled by a chromosome population. Next, each segmentation variant from the final population is tuned in an iterative process and the best chromosome is selected. Then character samples resulting from application of the segmentation modeled by the selected chromosome are grouped into sets corresponding to letters from the alphabet. Finally, the most outstanding samples are rejected so as to maximize the accuracy of words recognition obtained with a character classifier trained with the reduced samples set.

Segmentation-free recognizer is presented to transcribe Chinese handwritten documents, incorporating Gabor features and Hidden Markov Models (HMMs). Textline is extracted and filtered as Gabor observations by sliding windows first. Then Baum-Welch algorithm is used to train character HMMs. Finally, best character string in maximizing a posteriori criterion is found out through Viterbi algorithm as output. Experiments are conducted on a collection of Chinese handwriting. The results not only show the evident feasibility of segmentation-free strategy, but also manifest the advantages of Gabor filters in the transcription of Chinese handwriting.

Illegal trade and theft of coins appears to be a major part of the illegal antiques market. Image based recognition of coins could substantially contribute to fight against it. Central component in the permanent identification and traceability of coins is the underlying classification and identification technology. However, currently available algorithms focus basically on the recognition of modern coins. To date, no optical recognition system for ancient coins has been researched successfully. In this paper, we give an overview on recent research for coin classification and we show if existing approaches can be extended from modern coins to ancient coins. Results of the algorithms implemented are presented for three different coins databases with more then 10.000 coins.

In this paper we propose a new texture-based method for extraction of text areas in a complex document image. Gabor filter, motivated by the multi-channel filtering approach of Human Visual System (HVS), has been employed to create energy map of the document. In this energy map we assumed that text areas were rich in high frequency components. Connected components (probable text characters) were extracted by binarization of the energy map with Otsu’s adaptive threshold method. First non-text components such as pictures, lines, frames etc. were eliminated by Gabor filtering. As a novel approach, remaining non-text components were then eliminated by using character component interval tracing. Elimination that formed in two stage, enhanced the success of detecting text area on different kinds of digital documents.

Grey value thresholding is a segmentation technique commonly applied to tomographic image reconstructions. Many procedures have been proposed to optimally select the grey value thresholds based on the image histogram. In this paper, a new method is presented that uses the tomographic projection data to determine optimal thresholds. The experimental results for phantom images show that our method obtains superior results compared to established histogram-based methods.