Catálogo de publicaciones - libros

This paper presents a new robust approach for registration and segmentation. Segmentation as well as registration is attained by morphing of an N-dimensional model, the , onto the Ndimensional data. The approach is general and can, in fact, be said to encompass much of the deformable model ideas that have evolved over the years. However, in contrast to commonly used models, a distinguishing feature of the Morphon approach is that it allows an intuitive interface for specifying prior information, hence the expression . In this way it is simple to design Morphons for specific situations.

The priors determine the behavior of the Morphon and can be seen as local data interpreters and response generators. There are three different kinds of priors: – (elasticity, viscosity, anisotropy etc.), – (brightness, hue, scale, phase, anisotropy, certainty etc.) and – (filter banks, estimation procedures, adaptive mechanisms etc.).

The morphing is performed using a dense displacement field. The field is updated iteratively until a stop criterion is met. Both the material parameter and context fields are addressed via the present displacement field. In each iteration the neighborhood operators are applied, using both data and the displaced parameter fields, and an incremental displacement field is computed.

An example of the performance is given using a 2D ultrasound heart image sequence where the purpose is to segment the heart wall. This is a difficult task even for trained specialists yet the Morphon generated segmentation is highly robust. Further it is demonstrated how the Morphon approach can be used to register the individual images. This is accomplished by first finding the displacement field that aligns the morphon model with the heart wall structure in each image separately and then using the displacement field differences to align the images.

The aim is to compress and decompress structured volume graphics in a lossless way. Lossless compression is necessary when the original scans must be preserved. Algorithms must deliver a fair compression ratio, have low run-time and space complexity, and work numerically robust. We have developed PR0 to meet the goals. PR0 traces runs of voxels in 3D and compensates for noise in the least significant bits by way of using differential pulse-code modulation (DPCM). PR0 reduces data to 46% of the original size at best, and 54% on average. A combination of PR0 and Worst-Zip (Zip with weakest compression enabled) gives reductions of 34% at best, and 45% on average. The combination takes the same or less time than Best-Zip, and gives 13%, respectively 5%, better results. To conduct the tests, we have written a non-optimised, sequential prototype of PR0, processed CT and MRI scans of different size and content, and measured speed and compression ratio.

We consider lossless compression of digital contours in map images. The problem is attacked by the use of context-based statistical modeling and entropy coding of chain codes. We propose to generate an optimal context tree by first constructing a complete tree up to a predefined depth, and then create the optimal tree by pruning out nodes that do not provide improvement in compression. Experiments show that the proposed method gives lower bit rates than the existing methods for the set of test images.

Estimation of vectors is well-studied in estimation theory. For instance, given covariance matrices of input data allow to compute optimal estimates and characterize their certainty. But a similar statement does not hold for vectors and unfortunately, the majority of estimation problems arising in computer vision refers to such homogeneous vectors...

The aim of this paper is twofold: First, we will describe several iterative estimation schemes for homogeneous estimation problems in a unified framework, thus presenting the missing link between those apparently different approaches. And secondly, we will present a novel approach called IETLS (for iterative equilibrated total least squares) which is insensitive to data preprocessing and shows better stability in presence of higher noise levels where other schemes often fail to converge.

In image compression and feature extraction, linear expansions are standardly used. It was recently pointed out by Lee and Seung that the positivity or non-negativity of a linear expansion is a very powerful constraint, that seems to lead to sparse representations for the images. Their technique, called Non-negative Matrix Factorization (NMF), was shown to be a useful technique in approximating high dimensional data where the data are comprised of non-negative components. We propose here a new variant of the NMF method for learning spatially localized, sparse, part-based subspace representations of visual patterns. The algorithm is based on positively constrained projections and is related both to NMF and to the conventional SVD or PCA decomposition. Two iterative positive projection algorithms are suggested, one based on minimizing Euclidean distance and the other on minimizing the divergence of the original data matrix and its non-negative approximation. Experimental results show that P-NMF derives bases which are somewhat better suitable for a localized representation than NMF.

One of the key requirements for a cognitive vision system to support reasoning is the possession of an effective mechanism to exploit context both for scene interpretation and for action planning. Context can be used effectively provided the system is endowed with a conducive memory architecture that supports contextual reasoning at all levels of processing, as well as a contextual reasoning framework. In this paper we describe a unified apparatus for reasoning using context, cast in a Bayesian reasoning framework. We also describe a modular memory architecture developed as part of the VAMPIRE* vision system which allows the system to store raw video data at the lowest level and its semantic annotation of monotonically increasing abstraction at the higher levels. By way of illustration, we use as an application for the memory system the automatic annotation of a tennis match.

A novel method that is able to simulate artistic effects of ink-refusal and stroke-trace-reservation in ink paintings is developed. The main ingredients of ink are water, carbon particles, as well as glue. However, glue is not taken into account in other researches, although it plays an important role in ink diffusion. In our ink-diffusion model, we consider the number of fibers and the quantity of glue as parameters of the structure of paper. We simulate the physical interaction among water, carbon particles, glue, and fiber mesh of paper. The realistic renderings created from our models have demonstrated that our models are successful, and are able to imitate the special artistic effects of ink painting.

The spectral reflectance of Icons is estimated from RGB digital images taken by a digital camera, and it is applied to detect metameric color areas in the Icons. In this paper, two detection methods are proposed and examined by using a test chart and ten Icons painted on wooden plates. The first method is based on the definition of metamerism that two stimuli can match in color while having a disparate spectral reflectance. The second method is based on a phenomenon that the variation of the color difference between two colors is changed by replacing the illuminant if the colors are metamers to each other. The experimental results can be used to consider which parts of the Icons have been repainted as restoration treatments. Despite the necessity of further consideration and improvement, the experimental results demonstrate that the proposed methods have the basic ability to detect metameric color areas.

This paper describes an approach to digital archives of art paintings by considering the surface properties and the perceptual effect. A multi-band imaging system with six spectral channels is used for observing the painting surfaces. Multiple images of a painting are acquired with different illumination directions. Algorithms are presented for estimating the surface properties of surface normals, surface spectral reflectance, and reflection model parameters. All the estimates are combined for rendering realistic images of the painting under a variety of illumination and viewing conditions. Moreover, a chromatic adaptation transform is described for predicting appearance of the painting under incandescent illumination and producing the full color image on a display device. The feasibility of the method is demonstrated for an oil painting.

In this paper a novel method of spectral image quality characterization and prediction, preferential spectral image quality model is introduced. This study is based on the statistical image model that sets a relationship between the parameters of the spectral and color images, and the overall appearance of the image. It has been found that standard deviation of the spectra affects the colorfulness of the image, while kurtosis influences the highlight reproduction or, so called vividness. The model presented in this study is an extension of a previously published spectral color appearance model. The original model has been extended to account for the naturalness constraint, i.e. the degree of correspondence between the image reproduced and the observer’s perception of the reality. The study shows that the presented preferential spectral image quality model is efficient in the task of quality of spectral image evaluation and prediction.